AIRCRAFT SCHEDULING:
AN INTERACTIVE GRAPHICS APPROACH
by
BRUCE CURTIS LUBOW
S.B., Massachusetts Institute of Technology
(1979)
SUBMITTED IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE
DEGREE OF
MASTER OF SCIENCE
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
January 1981
) Massachusetts Institute of Technology 1981
Signature of Author
----
-
Department of Aeronautics and Astronautics
January
13, 1981
Certified by
Antonio L. Elias
X
:~
X
Thesis SuperVisor
Accepted by
MASSACHUSETTS
INSTITUTE
OF TECHNOLOGY
FEB
2 0 1981
LIBRARIES
Harol Y. Wachman
Chaiffman,Departmental Graduate Committee
-2-
AIRCRAFT SCHEDULING:
AN INTERACTIVE GRAPHICS APPROACH
by
BRUCE CURTIS LUBOW
Submitted to the Department of Aeronautics and Astronautics
on January 16, 1981
in Partial Fulfillment of the Requirements for the Degree
of Master of Science
ABSTRACT
The problem of scheduling aircraft is approached with the intent of
providing a computerized tool to play a central role in the schedule development
process. It is determined that automated approaches and batch processing aids
provide supplementary assistance but that interactive graphics are essential if the
manual scheduling methods currently employed are to be replaced and improved
upon.
Various visual representations
of schedule data are reviewed and the
sequence chart is determined to provide the best choice for mechanization in
terms of usefulness and practicality.
Designs are presented for various scales of
the basic sequence chart and a set of variations that are determined to provide
the information required by schedulers to make each of the major decisions
involved in the scheduling process. Those displays are designed for the possibility
of implemention on terminals with color, special character sets and highlighting
features.
A set of commands is presented which utilize full screen panels with
restricted cursor movement to guide the user through data entry processes.
Commands are included to produce and control the content of the various displays
and to enter and manipulate schedule data.
Thesis Supervisor:
Antonio Elias
Title:
Assistant Professor of
Aeronautics and Astronautics
-3-
ACKNOWLEDGEMENT
I am indebted to Professor Antonio Elias as my advisor on this project for
My sincere gratitude is
numerous helpful suggestions and constant support.
extended to William Pacelli of Eastern Airlines and Jerry Frisora and John Scalea
of USAir for the information they supplied on the status of scheduling at their
companies. I am also indebted to Professor Robert Simpson for introducing me to
this topic and encouraging me to pursue it.
Acknowledgement is also made to Ruth Erickson and Linda Martinez for
their dedication and extraordinary typing competence.
-4aAIRCRAFT SCHEDULING:
AN INTERACTIVE GRAPHICS APPROACH
Table of Contents
PAGE
i.
INTRODUCTION
2.
THE SCHEDULING PROBLEM
2.1
2.2
3.
The Makings of a Valid Schedule
12
2.1.1
Resources
13
2.1.2
Components
14
2.1.3
2.1.4
Relationships
Constraints
17
19
Desirable Qualities of a Schedule
2.2.1
The Customer's Requirements
23
2.2.2
The Airlines' Viewpoint
27
CURRENT MANUAL AND COMPUTERIZED
3.1
23
Visual Representations
3.1.1
3.1.2
3.1.3
SCHEDULING AIDS
of Schedules
32
32
Two Dimensional Projections
Cross Section Views of Schediile
34
Maps
41
Projections of Schedule Maps
47
3.2
Existing Computer Aids for Scheduling
53
3.2.2
Automated Programs
Batch Processing Aids
3.2.3
Online Scheduling Aids
56
3.2.1
4.
INTERACTIVE GRAPHICS DESIGN FOR SCHEDULE DEVELOPMENT
4.1
52
54
61
The Advantages of Interactive Graphics
61
4.1.1
62
Simplification
-4bTABLE OF CONTENTS (Continued)
PAGE
4.1.2
4.1.3
4.2
4.3
4.4
Visualization
Communication
Information Requirements of the Schedule
Development Process
4.2.1
Production 68
68
4.2.2
Manipulation
70
4.2.3
Validation 74
4.2.4
Evaluation 77
Requirements of CRT Displays
80
4.3.1
Size
80
4.3.2
Resolution
4.3.3
Format Generality
83
Displays Design
4.4.1
4.4.2
4.4.3
5.
85
87
Level One: The Basic Sequence Chart
Display
4.5
64
65
87
Level Two: Modified Displays
100
4.4.2.1
Scale Modifications
100
4.4.2.2
Rearrangement
115
4.4.2.3
Highlight
Level Three: Combined Modifications
126
129
Interactive Commands and Working Panels
4.5.1
Access Sieve
143
144
4.5.2
Display
152
4.5.3
Access Flight
156
4.5.4
4.5.5
Sequence
Crossover
168
168
4.5.6
Access Global Parameters
171
RECOMMENDATIONS
BIBLIOGRAPHY
AND CONCLUSIONS
172
175
-5'aLIST OF FIG URES
PAGE
FIGURE 1-1
Organizational
FIG URE 3-1
Three Dimensional Schedule Space
33
FIG URE 3-2
Route Maps
35
FIG URE 3-3
Typical Schedule Map
39
FIG URE 3-4
Station Activity Representations
44
FIG URE 3-5
Routing Charts
48
FIG URE 3-6
Typical Sequence Chart
51
FIG URE 4-1
Three Levels of Display Derivatives
88
FIGURE 4-2
Alternative Data Arrangements
92
FIG URE 4-3
Complete Sequence Chart Display
96
FIG URE 4-4
Use of Background Color in Sequence Display
98
FIG URE 4-5
Expanded Sequence Chart Display
102
FIG URE 4-6
Compressed Display Scales
103
FIG URE 4-7
Altered Format Compressed Display Scales
105
FIG URE 4-8
Origination and Termination Display Format
108
FIGURE 4-9
Daily Sequence Summary of Originations
and Terminations, Flight Numbers and
View of Scheduling
7
Supplemental Information
109
Daily Sequence Summary of Flight
Numbers and Supplemental Information
112
FIG URE 4-11
Daily Sequence Summary of Flight Numbers
113
FIGURE 4-12
Sequence Inventory
114
FIGURE 4-10
-5bLIST OF FIGURES (continued)
PAGE
FIG URE 4-13
FIG URE 4-14a
FIG URE 4-14b
FIGURE 4-14c
FIG URE 4-15a
FIG URE 4-15b
FIGURE 4-16
FIGURE 4-17
Aircraft Type Inventory
Sequence Display for Turns to Specific
Flight
118
Sequence Display for Turns by Station
and Time
119
Sequence Display for Turns in First-In-
First-Out Order
120
Sequence Display for Crossovers to
Specific Flight
122
Sequence Display for Crossovers by
Station and Time
123
Ramp Chart Display
125
Ramp Chart with Incremental Aircraft
Count Highlight
FIG URE 4-18
132
Compressed Turn Display for Single Time
Band
FIG URE 4-19
116
134
Compressed Turn Display for Multiple
Dates
136
Compressed Crossover Display for Single
Time Band
137
FIG URE 4-21
Compressed Crossover for Multiple Dates
138
FIG URE 4-22
Compressed Station Activity List
140
FIG URE 4-23
OAG Timetable Service List
142
FIG URE 4-24
Access Sieve Working Panel
151
FIG URE 4-25
Initial Access Flight Working Panel
158
FIGURE 4-26
Typical Completed Access Flight Working
Panel
161
Sequence Working Panel
169
FIGURE 4-20
FIGURE 4-27
-5cLIST OF TABLES
PAGE
TABLE 4-1
Character Definitions
TABLE 4-2
Command Language Symbols and Syntax
145
TABLE 4-3
Display Options Menu Items
155
99
-6-
1.
INTRODUCTION
To the casual observer airlines seem to be on the leading edge of high
technology applications.
They use modern and sophisticated aircraft,
a highly
advanced reservation system and communications network, and the most refined
navigation and electronics equipment available.
So it is disillusioning to discover
that in the back rooms, corporate planning, and especially aircraft scheduling, has
not kept pace with recent advances in modern techniques and technologies.
This is even more surprising considering
function is to the success of an airline.
how critical
the scheduling
Although the scheduling of manpower,
production and inventory resources is performed in many industries, it rarely
assumes the significance that it does for an airline.
In most industries the
selection of products to produce, the methods of production and the markets to
target are somewhat independent functions.
however, combines all of these functions:
The scheduling process of an airline,
the schedule is at one time the means
of production, the product and the marketing plan.
Scheduling plays such a central role because it is the sole output and
objective of the airline.
All departments, personnel and resources ultimately
contribute to the determination
and implementation of the schedule.
Figure 1.1
shows how scheduling interacts with other functional groups within a typical
airline.
-7-
oL
U,
?:a
W
=
Z
-
a
-8-
No function in the airline can be performed independently of the schedule.
Each department
must provide the schedulers with an estimate of resources and
requirements and receive similar feedback from them. Scheduling coordinates the
activities
of the airline by supplying a focal point for the planning process.
Because of this coordinating role, scheduling occupies one of the highest positions
in the management hierarchy.
The planning cycle is normally a hierarchical activity
with each level's
decisions influencing the level below to a much larger extent than the one above.
Airline planning, however, does not fit this rigid methodology since scheduling
cuts across all three levels.
Essentially all corporate level planning, from fleet
planning to mergers, must be evaluated in terms of its impact on the schedule.
Each alternative must be quantified in terms of a detailed schedule, which can be
further evaluated, before a decision can be made. Only limited confidence exists
in aggregate methods for forecasting the impact of decisions because complex
issues in the scheduling problem often make otherwise reasonable management
goals impracticable.
In spite of its importance, scheduling is still done by a few analysts using
colored pencils and large pieces of paper taped to the wall. Attempts have been
made to mechanize some of this process but have met only limited success.
Algorithms to automatically generate schedules have only been able to deal with
limited subsets of the total problem, and have therefore not had a major impact
on the scheduling function and programs to provide batch processing services have
only had a peripheral influence on scheduling. On line access to data bases and
other schedule development tools, while steps in the right direction, have not yet
reached the level of sophistication required to directly aid the scheduler.
-9-
This analysis proposes that a computerized system that can handle the full
complexity of realistic commercial schedules, provide interactive
response to
human decisions and utilizes graphics capable of replacing and even improving
upon the wall charts now used, is the appropriate tool to improve the quality of
schedules and the speed at which they are developed.
The availability
of
moderately priced stand-alone computer systems with full graphics capability and
the realization that the human role in decision making processes as complex as
scheduling can not be eliminated, have made this approach the most promising at
this time.
The need to improve the scheduling process is particularly strong at this
time.
Increased fuel prices, and competitive pressures created by deregulation,
can best be dealt with through better and more responsive planning. Because of
its significance to airline planning, scheduling is probably the best candidate for
improvement.
By increasing the quality and speed of schedule development, an
airline is equipped to investigate more viable alternatives in each planning cycle.
The following analysis identifies
schedule and then presents
the elements
involved in building a
a design for a set of displays and interactive
commands that will allow a schedule to be developed entirely on a computer. The
computer tool presented here is designed to supplement schedule development,
not to automate it.
This tool is intended to replace the manual methods now
employed, and, in so doing, provides a wide range of improvements in the way the
complex schedule data base is produced, recorded, modified and displayed.
Implicit in the approach described is the assumption that the human role is
indispensable in the schedule development process.
All aspects of the design are
-10-
intended to contribute to easing the scheduler's job by taking over routine
deterministic tasks, presenting information in the most convenient form possible
and minimizing the effort required to control the computer. The goal is to leave
the scheduler free to make important decisions and to present the information
essential to making them.
Section 2 provides a background on the scheduling problem including
definitions of the important terms and concepts needed to understand subsequent
discussions.
Section 3 reviews the many manual graphic aids currently used by
schedulers in industry today as well as a number of the existing computerized
tools designed to aid this process.
of interactive
In Section 4 the advantages and requirements
graphics are discussed and designs for reference
active displays and a controlling command language are presented.
displays, inter-
-11-
2.
THE SCHEDULING PROBLEM
The complexity of the aircraft scheduling problem has best been described
by Melvin Brenner of American Airlines as "...trying to put together
puzzle, constructed
in three
a jigsaw
dimensions, while the shape of key pieces is
constantly changing." (Tipton, Stuart G., et al. 1961). A formidable task indeed.
This analogy is an excellent one, but the problem is, in fact, even more
complex than this might indiciate.
A jigsaw puzzle is normally made up of a
single material such as cardboard.
These pieces are two dimensional and of a
constant shape, the interlocks between pieces are unvaried both in number and in
kind, and finally, the precise picture printed on the pieces is not only unchanging
but is known in advance.
Scheduling, however, has none of these consistencies.
Many additional
degrees of freedom exist for each aspect of the puzzle. A variety of materials
are used to construct
three dimensional pieces of flexible size and shape.
A
number of interlocks are required but nearly an infinite number of solutions exist,
not all of which are equivalent.
Other interlocks may optionally be specified to
alter the final picture.
This picture must meet certain criteria for quality and excellence, and
must have certain characteristics and some specific images present, but the
actual final picture is the scheduler's decision. The scheduler has a conception of
the image he hopes to produce, he has his colored pencils and an understanding of
the forms and relationships that the schedule may contain. The process begins
-12-
with certain portions that he is confident belong, and proceeds with other
attractive additions.
Some aspects will augment each other and bring the
schedule to life, other portions will be overwritten and drawn again to fit the
emerging patterns.
In the following sections each of these aspects of scheduling will be defined
and expanded. First, some of the major resources that are used, the construction
of the components of the schedule or puzzle pieces, the nature of the interlocks,
and the requirements and characteristics
of a completed schedule are all defined.
Next, some of the techniques for visualizing this complex three dimensional
problem in two dimensional media are discussed.
Finally, the qualities
and
characteristics that constitute the goals of the scheduler or the desired picture
that the finished schedule should depict are described.
2.1
The Makings of a Valid Schedule
To understand the problem the various elements that make up a schedule
and the relationships between them must be precisely defined.
discussion presents these key definitions and concepts.
The following
Unfortunately, these
terms have not been used in a consistent and rigorous manner in practice,
so to
clarify and delineate these concepts each term will be given a precise definition,
even though some of them will not conform to standard usage.
-13-
2.1.1
Resources
Scheduling is concerned primarily with two basic resources:
airports.
airplanes and
Without either of these there would not be air transportation
or any
need to schedule air transportation services.
Within these categories a number of distinctions can be made. An aircraft,
as used here, refers in a generic sense to any "aircraft of a particular type."
An
aircraft type is a specific configuration of technical and performance characteristics such as:
identifier
model, seats, range, speed, etc.
such as 727-200, DC9-30, etc.
Associated with these is an
When referring
to a single specific
aircraft of a given type and with a unique tail number identifer such as A1234B,
the term piece-of-equipment
will be used. Finally, an aircraft fleet is a number
of pieces of equipment of the same type which can be used interchangably.
The scheduler is concerned with a fleet or a number of fleets of aircraft
that can be flown between airports.
corporate
Although it is of some importance to the
planning process, and even to a large degree to the scheduler,
the
details of the operation of this fleet are ignored in this definition of the
scheduling problem.
So for now, this fleet is assumed to be manned with flight
and cabin crews, fueled and ready to go.
Airport
Again, the scheduling problem will be initially defined without any consideration of the details of station operations.
Each station will be presumed to have
sufficient gates, man power, fuel, baggage and passenger handling facilities, etc.
-14Components
2.1.2
The basic component, or building block, of a schedule is the flight. A flight
is a complex combination of the resources forming a relationship in both space
and time.
Each flight is identified uniquely by its flight number, currently a
combination of up to four alphanumeric characters.
A unique aircraft
type is
associated with a flight and, eventually, one or more pieces of equipment will be
assigned to a flight.
The construction of a flight begins by joining pairs of stations together to
form links over which aircraft
will fly.
A whole series of links on which
passengers can travel in succession is known as a routing. Links and routings are
space dimension variables which can be combined with specific departure
arrival times.
segment.*
and
When a link is supplied with these operating times it becomes a
Similarly, a routing is transformed into an itinarary when the time
dimension is included.
In addition to the operating times, itineraries contain two types of event
duration times: block time, which is the duration of a segment from the time the
aircraft leaves the gate at one station to the time it docks with the gate at the
next station; and, transit or station transit time, which is the time between
successive segments of an itinerary.
These four times (Departure, Arrival, Block
*The term "segement" is frequently used to refer to a link in which adds to the
confusion of this terminology.
-15-
and Gate) can be used in combination
to unambiguously define
a flight's
operations, however, under or over specification is possible and must be avoided.
Proper specification requires that the following conditions be met:
1)
At least one arrival or departure time must be explicitly specified
for each flight itinerary.
2)
Exactly one block or gate time between two specified arrival or
departure times must be left unspecified.
3)
All block and gate times before the first specified departure time
and all block and gate times after the last specified arrival time
must be explicitly specified.
A flight, then, is composed of a number of itineraries.
be applicable
application.
to a flight on certain
Each itinerary may
days, this is known as the itinerary's
Normally, an application will begin on some date, continue until a
later date and will be valid on certain days of the weeks during that period.
A
single flight may only contain one valid itinerary on any given day although it
need not have an itinerary valid on every day.
In addition to being a collection of aircraft movements between stations at
specific times, a flight also specifies a set of markets to be served.
These
services are of a variety of types and quality and may not correspond directly to
the segments of the flight. Three types of service are possible : nonstop, direct
or through, and connecting.
Each of these provides transportation
between one
market area where a passenger has intentionally spent time, known as his origin,
-16-
and a second market area where the passenger will also remain for some time,
known as his destination.
a)
Non stop service is provided by a single segment of a flight.
It
allows a passenger to travel between his origin and destination on a
single plane and without intervening stops.
b)
Direct or through service provides a link between two market areas
using more than one successive segments of a single flight.
This
allows a passenger to travel between his origin and destination on a
single plane but making one, two or more stops enroute.
c)
Connecting service is provided by segments on more than one flight.
This service carries a passenger from his origin to an intermediate
station, known as a connecting point, where the passenger can take
another flight to his destination.
Service can be provided with a
combination of through and connecting flights.
Connections be-
tween two flights provided by the same airline are known as online
connections, as opposed to interline connection between flights on
different airlines.
A single flight of N segments can provide N nonstop services, N-1 one stop
services,
N-2 two stop services,
etc.
In addition, an unlimited
number of
connecting services could be provided at any or all stations in the itinerary.
However, it is only desirable to provide a selected subset of the possible services.
The ability to exclude services from flights adds another degree of freedom to the
shape of the flight.
-17-
Therefore,
the three primary components of a schedule are:
timing, and service exclusion.
great deal of flexibility.
independently.
routing,
Each of these is variable allowing the scheduler a
On the other hand, none of these can be specified
The services possible depend on routing and, in the case of
connections, on timing. Timing, of course, depends on routing and services to be
offered and routing is similarly dependent on the other two.
2.1.3
Relationships
Once a number of flights have been created the process of joining the
pieces together can begin. As this process proceeds new flights can be added and
the old ones rearranged to meet the scheduler's needs. Rather than making the
problem easier to solve, this degree of freedom increases the viable options open
to the scheduler and lengthens the search for a better solution. Still, to a large
extent the scheduler can manipulate the pieces of the schedule independently
while retaining the option of returning to the flight creation phase to improve the
shape of the pieces.
Flights can be joined together into larger blocks of the schedule. However,
this is not an easy matter, since two different types of interlocking relationships
between flights are possible, and these interlocks must be compatible in both the
space and time dimensions.
The first interlock between flights is known as the turn. A turn indicates
the path that an individual aircraft will take in flying a series of flights.
Thus, an
aircraft that flies one flight may then turn to another flight and then another and
-18so on. A string of flights so joined will be called a sequence, although the term
"cycle" is often used for this concept.
A flight can only turn to another flight if they are compatible in both time
and space as well as equipment type.
terminate:
It is insufficient for the first flight to
an origination must occur at the same station as well. The ground
time constraint is usually extended so that a specified interflight or turn time is
allowed between two flights turning at a station.
As discussed earlier, a flight is not a single itinerary nor is it likely to be
flown on only a single day. Each application of a flight is a seperate unit which
can be turned to different flights on different days. There is no requirement that
applications of a flight on different days be members of the same sequence. Thus,
not only are flights three dimensional but they are discontinuous with parts at
many times and locations.
All of these parts must be considered when forming
the pieces and interlocking them.
Sequences can continue for days, weeks or months. Each sequence is flown
by one aircraft
independent
throughout
this period.
schedule with no relationship
A sequence can be a complete
to other sequences;
and
however, this
situation rarely exists. Rather, sequences are linked together using the second of
the two interlocking methods.
Crossovers also join flights in both the space and time dimensions. A cross
over is the relationship between two flights, one arriving at a station before the
second one departs.
Again the times must be separated by a specified time known
as the connection
time.
The crossover
produces a number of connection
-19-
possibilities for passengers on the arriving flight. This distinction between crossovers-the
occur-and
logical relationship between flights that permits connections to
connections themselves, must be understood.
Flights may act as both
donors and recipients of traffic flow if the crossovers are provided both ways
between flights.
Many flights may be crossed over to one another at a single
station during a period of time.
These arrangements
of many flights mutually
exchanging traffic are known as connecting banks or connecting complexes.
Thus, flights may be joined together both into sequences and connecting
complexes.
However, the same set of flights must be joined into sequences and
linked into one or more connecting complexes making these processes highly
interdependent.
The flights must be so configured to permit as many desirable
crossovers to occur as is possible while still permitting all the necessary turns to
occur.
2.1.4
Constraints
Many limits to the freedom to shape and join the different
aspects of
scheduling exist. These constraints affect both airports and aircraft and arise
from both internal economic causes as well as external regulatory ones.
Of
course, many of the constraints are three dimensional, specifing simultaneous
space and time limits.
Fleet size is probably the most important and immediate concern of the
scheduler.
Although aircraft can be leased and purchased given sufficient time,
the scheduler rarely has this flexibility. The first implication of this constraint is
-20-
that no more simultaneous sequences can be scheduled than there are aircraft
available of the type specified to fly those sequences.
Of even more signficance is the constraint that the fleet size must at least
remain constant during a schedule period. Therefore, at the very least, exactly as
many aircraft arrive at each station as depart that station.
constitutes
balanced flow.
This condition
If this condition were not met, stations would be
acting as aircraft "sinks" and "sources."
Aircraft flow can be balanced over any
period of time; for instance, each station may be balanced every day or only once
a week.
A network with balanced flow may have continuous flow. Continuous flow
exists when every flight turns to another flight so that sequences have no gaps
from the beginning to the end of a schedule period.
Continuous flow can exist
over a period without resulting in a net balanced flow over the entire period.
Thus continuous flow could carry a fleet of aircraft from some initial geographical distribution to another one over an extended period of time.
If a network has both continuous and balanced flow then it is said to have
cyclic flow. A cyclic schedule not only maps a continuous flow of aircraft but
ends in a state identical to its initial state thereby permitting the schedule to be
repeated.
Individual sequences can repeat at regular intervals, for instance daily,
and are thus themselves cyclic. This is the origin of the term cycle which is often
used to refer to any daily sequence whether it is cyclic or not.
Maintenance restrictions
impose another significant constraint on sche-
dules. Routine maintenance that must be performed on a regular basis can only
-21be accomplished at the airlines maintenance bases. This requirement means that
every sequence must include a turn at one of the airlines maintenance bases of an
adequate duration at regular intervals.
For a typical airline (e.g., USAir), this
means that aircraft must spend every other night at a maintenance base.
Often, resequencing flights can provide adequate maintenance time for all
aircraft without rerouting any flight itineraries. Flights can also be ferried for
maintenance,
however, careful planning can usually eliminate the need for this
wasteful operation.
This planning includes locating maintenance bases as well as
scheduling aircraft.
A parallel
problem to the scheduling of aircraft
is the scheduling of
aircraft crews. Although not strictly a constraint, inattention to crew requirements
and limitations
can result
in inefficient
schedules.
Because of both
regulatory and union rules concerning flight and cabin crew, many constraints are
placed on their utilization.
Limits are placed on actual flight time, duty time
which includes the non flight portion of the crew's job, and overnights which result
when aircraft schedules and duty time restrictions prevent the crew from
returning to their home base for a required rest period. These constraints on crew
utilization translate into costs for the airline when schedules are so configured as
to require excessive overnights, deadheading, and other inefficiencies. Although
not the primary task of the scheduler, this must be a consideration.
An important set of constraints involves the capacity of individual stations.
These take four forms: gate capacity, station staff capacity, fuel availability,
and regulated airport access (slot) capacity. Again these are combined time and
space constraints. The average capacity of a station is rarely a problem, rather it
-22-
is the peaking of activity at busy times of day that overloads the fixed resources
at the station.
A scheduler must pay careful attention to the peaks of activity to
keep them within acceptable levels.
This is particularly true for airports with regulated access capacity which
must be strictly complied with. Four U.S. airports (DCA, ORH, JFK, and LGA)
have already begun setting fixed limits on the number of operations (takeoffs or
landings) at the airport per hour, known as slots. Other airports, such as DEN, are
close to ths situation.
Curfews also exist at many airports restricting the
operating times of the earliest and latest flights of the day.
Another critical situation for station activity peaking occurs at small or
infrequently used stations. It is bad enough that facilities must be maintained for
the handful of flights that use them; even more inefficiencies result if a schedule
calls for two aircraft
to be serviced simultaneously while station facilities and
staff remain idle throughout the remainder of the day, or when flights are spaced
so far apart that a second shift must be hired to cover both.
Another restriction on the use of airports relates to the types of aircraft
permitted to operate there.
This limit can arise from noise regulation, which may
vary by time of day, or from runway length or construction, or from gate size.
The quantity of fuel available at a station may also restrict the number of flights
or the aircraft types that can stop there.
-23-
2.2
Desirable Qualities of a Schedule
The previous section described the materials, pieces, interlocking relationships and constraints involved in the schedule problem.
and characteristics
of a "good" schedule?
Clearly,
possible, but not all of them are equally attractive
But what are the qualities
many valid schedules are
to the customer or the airline.
In addition to meeting the many constraints, a scheduler attempts to
produce a plan that constitutes the best possible compromise between the
requirements of the customer and the goals of the airline. This section describes
who some of the customers are and their needs, as well as some of the desirable
qualities of a schedule from the standpoint of airline management.
The Customer's Requirements
2.2.1
The Importance of Service/Demand to Scheduling
As in any industry, the airline's customers are unconcerned with the means
of production.
Their primary concern is the resulting product.
The product, of
course, is transportation between two points, but there are many other aspects
that are of interest to the customer.
Furthermore, there are many different
types of customers, each desiring not only service between different points but
also at different times and with a variety of other features.
Customers are attracted by a variety of factors unrelated to the actual
schedule.
These include onboard meals and entertainment,
convenient reser-
-24-
vations and ticketing procedures, fast and reliable baggage or cargo handling,
attractive and comfortable terminals and aircraft, and many others. For the
purposes of this discussion, only those considerations that are influenced directly
by the schedule will be considered.
As far as the customer is concerned, the schedule is what is presented in
the Official Airline Guide (OAG) and similar timetable publications.
He is not
concerned with how aircraft are routed and turned nor with the complex
compromises that resulted in producing crossovers, satisfied station constraints or
economical crew duty rotations.
The bottom line is the table of services between
a passenger's origin and his destination.
To understand the customer's viewpoint it
is only necessary to study how he will perceive various alternative
services as
presented in the OAG or indirectly through a travel agent or airline reservations
agent.
The market for air transportation
is a diverse and highly heterogeneous
one. Both passengers and cargo are carried by airlines.
Demand exists between
many market areas, which are defined as geographical regions that are homogeneous in that demand for service to them is not affected
by which airport
located there is used. A market is a directional pair of market areas such as Los
Angeles to New York or vise versa. Demand for transportation
in each market is
highly variable with the time of day, day of the week and season of the year.
Each of these dimensions of demand combine to produce a multiplicity
of
different components to the total demand for transportation services.
Passenger traffic has traditionally been partitioned into two classes:
business travelers and pleasure travelers.
The business travelers in each market
-25are highly sensitive to the time of day and day of the week on which flights are
offered.
They are most likely to travel during traditional commuting periods so as
to be able to conduct business on the same day as their trip, or so as to reduce
interference
with their working and rest periods.
In short and medium haul
markets they usually prefer to travel on week days, whereas on longer haul,
international
routes,
they often choose to travel on Sunday so as to have
sufficient time to get settled and rested before Monday morning. (See Elias,
1979).
On the other hand, pleasure travelers are more likely to travel during the
less hectic mid-day hours and on weekends.
seasonal variations
in the
Much larger differences
in demand exist for pleasure travelers
than for business
travelers.
Cargo demand tends to be higher at night. (Taneja, 1979)
Various
companies, freight forwarders and the post office collect cargo throughout the
day to ship after normal business hours for delivery before the start of the next
business day.
Concern for flight times involve a combination
departure times.
of local arrival
and
This is of particular concern for flights crossing a number of
time zones. A five hour flight departing from Los Angeles at 6:00 PM will arrive
in New York at 2:00 AM, clearly an inconvenient time for most travelers.
problem alone causes very different
patterns
markets depending on their length and direction.
of demand between
This
different
-26The type of service offered is another aspect of the schedule affecting the
customer.
It is not merely the increased time of a through or connecting path
that disturbs passengers and cargo shippers.
delays and are uncomfortable for passengers.
Stops increase the probability of
Connections create the possibilities
of missing a flight and losing baggage or cargo in addition to the drawbacks of the
stop involved. Non-stop service is always the customer's first choice:
however, a
through flight or connection at a more convenient time may outweigh inherent
disadvantages.
The number of competing
nonstops in a market
and their
departure times will determine whether other services will be acceptable.
Finally, flight times as close together as one minute may not be viewed
equally by the customer.
This condition arises from a variety of causes.
Since
services are listed in chronological order the earliest one is the first one
encountered and often the one chosen for that reason alone.
The other two
considerations for setting flight times are related to passenger convenience.
Passengers making frequent trips become accustomed to using particular services
at specific times.
service
They may have worked out a routine around a particular
that they are comfortable
with and don't wish to change.
convenience is regular service such as hourly, daily, etc.
Another
Also, service times that
are round numbers are easy to remember and more appealing so an 8:00 flight
might be chosen over a 7:53 flight just for the convenience of on-the-hour service.
If service were the only consideration,
independently
of aircraft
routings that
a timetable
might be a highly desirable product.
However, only the objectives for a desirable timetable
schedule must be constructed to produce a timetable
one as possible.
could be drawn up
can be specified and a
as close to this perceived
-27-
2.2.2
The Airline's Viewpoint
In trying to produce the desired timetable from the service standpoint,
while meeting the many constraints,
the airline is also trying to maximize its
revenue and minimize its costs. Although these goals are often contradictory,
carefully planned schedules will optimize potential earnings.
A variety of routing and timing decisions are influenced by the goal of
efficient use of the airline's resources. The primary concern of the scheduler is to
provide flights that will produce sufficient volumes of traffic to fill the aircraft
to profitable levels. This goal can be accomplished through a variety of means.
Some markets, because of their high density, geographic location or heavy
competition will easily justify nonstop service.
In many markets, however, the
frequency of flights needed to meet competitive
schedules and to provide an
adequate level of service exceeds the quantity justified by the traffic in that
market
alone.
Two approaches
can be taken
in this situation
to provide
reasonable service without committing the airline to an unprofitable situation.
Both of these involve combining traffic flows from multiple markets on single
flight segments. Which technique will be used depends primarily on the geography
of the network.
The first of these techniques is sometimes called cascading of traffic. This
involves creating aircraft routings between complimentary areas. For example,
imagine a hypothetical
set of four cities, A, B, C, and D.
If these were all
-28connected to one another by nonstop service the number of separate segments
required to serve each market once would be N(N-1), where N is the number of
cities; or in this example, N(N-1) = 12 segments.
But if these cities could be linked together by a single route from A to B to
C to D flown once in each direction, then service would be provided in each
market with only (N-I), or 6 segments.
If a link is added between D and A to
route the aircraft in a closed loop, then only N, or 4 segments are required.
In
addition to saving the costs of flying the extra segments, each flight will now
have passengers from 3 different markets on each segment.
flight will carry AB, AC and AD traffic.
Between A and B the
Between B and C it will carry AC, BC,
and BD traffic, and so on.
This simple routing has simultaneously reduced the cost of serving those
markets while increasing the traffic on each segment that will be flown. Of
course, the quality of service has declined from nonstop to through service in 6 of
the 12 markets.
For this reason and for a number of others that will be discussed
later, this technique must be used with a great deal of care. Still it is a good way
to provide economical service
in markets
in which it would otherwise
be
impossible.
Another opportunity arises to provide economical service in low density
markets that share a common origin or destination. An isolated market area may
not produce enough traffic in any single market to justify service.
But all traffic
originating at this station can be combined and carried as a whole to a convenient
connecting hub. At this point the aggregated traffic flow can be parceled out to a
bank of flights departing that station for any number of destinations.
This
-29procedure also helps to cascade traffic on those departing flights by increasing
the traffic using various segments and combinations of segments on these flights,
at the expense of making some, if not most, of the direct services into connecting
services.
Despite the power of aggregating traffic
flows, many drawbacks exist.
While very high load factors may be achieved using these techniques, profits
might actually suffer if they are not used properly.
A scheduler must be careful
to consider costs and revenue above mere traffic.
To really maximize airline profitability not only must high load factors be
achieved on most flights, but also the utilization
(average number of hours an
aircraft is flying per day) must be high. High equipment utilization is essential if
the airline is going to be able to cover its fixed costs.
Utilization is limited by
the ability of the airline to find markets with sufficient demand. At less popular
times utilization must be sacrificed to preserve average load factor.
The length of a flight segment,
important impact on profitability.
vantages.
known as stage length, also has an
Short stage lengths have two major disad-
First of all, the high costs of takeoff and landing, including ground
handling of the aircraft, passengers, baggage and cargo constitute a large portion
of the segments cost.
Although fares can be higher per mile on short haul
segments, they cannot be raised indiscriminately without forcing traffic into
other modes of travel which are readily available for shorter trips. In addition, on
short stage lengths the gate time between segments is a larger portion of the
block time. Thus the maximum utilization possible is severly reduced.
-30-
Yet the objectives of cascading and combining traffic flows is inconsistent
with the objective of maximizing stage lengths.
Multiple stops and connections
require shorter stage lengths than do nonstop links between distant stations. Not
only are shorter segments being flown, thereby increasing costs, but, in addition,
long haul traffic that is occupying space on many short haul segments cannot be
charged for the added costs of the stops or the added circuity
(see below).
If
long haul passengers make up a large portion of the traffic on these short haul
segments,
service
could be improved and costs
reduced by providing more
nonstops.
Another related problem is that of seat "block out". This occurs when a
flight contains both short and long haul segments and there is significant demand
for service in the short segment. On such a flight it is particularly desirable to
fill up the long haul segment or segments since these will produce the most
revenue.
If one of the short haul segments becomes filled primarily by local
traffic, then traffic between stations upline from this blocked segment that would
have continued on through the long segment cannot use the flight.
For example,
imagine a flight from A to B to C to D, where BC is a long haul segment and AB
and CD are short hauls. If AB demand is high, then AC traffic, which would help
to fill the important BC segment, will be blocked out.
One other factor must be considered when combining links to form routes.
This is the issue of circuity. Fares, at least standard coach fares, are normally
based in large part on the great circle distance between the origin and destination
stations. The actual path taken including intermediate stops and connections may
be much longer, or circuitous, than a direct flight. This not only results in poorer
service to the customer, but also dilutes the revenue per mile, or yield, received.
-31In conclusion, the ability to cascade and connect traffic is a very powerful
tool that must be used with care.
It is possible for inappropriate
routings and
connections to dilute fares, block out traffic and reduce utilization. On the other
hand, proper combinations of the schedule elements can produce a tremendous
synergy.
Each flight can be both receiving and contributing traffic support to
many other flights in a profitable manner.
All elements of the network can work
together in consistent and compatible ways that result in far greater output than
the individual flight segments could conceivably produce alone.
-323.
CURRENT MANUAL AND COMPUTERIZED SCHEDULING AIDS
3.1
Visual Representations
of Schedules
How can a scheduler expect to grasp the immense complexities of the
three dimensional and ill-defined aspects of the scheduling problem?
Ways have
been found to represent the elements and relationships of a schedule so that they
may be more readily understood and manipulated.
visual or graphic models of the complete
These representations
three dimensional picture
are
of the
schedule.
What could have been a four dimensional problem (three in space plus time)
is conveniently reduced to three dimensions by ignoring the vertical component altitude.
Thus the problem is one of motion on a two dimensional plane or map.
The vertical
axis in a three dimensional drawing of a schedule is now free to
represent time. Aircraft movements can be depicted as space-time lines winding
their way from station to station while continuously progressing through time.
Figure 3-1 shows a sketch of such a representation for a single aircraft sequence.
While this representation accurately depicts all of the events that make up
a schedule,
it is not a practical
tool.
The difficulties
of drawing three
dimensional relationships on a two dimensional medium make this method cumbersome. This will be even more true as more aircraft sequences are added. As the
event lines crisscross through the network the diagram will become increasingly
difficult
to read.
A mechanical version of this model can be imagined using
strings to represent sequences that wind between posts representing each station.
This would soon produce a tangled web of interwoven sequences that would be
impossible to interpret.
-33-
TIME
EAST
SOUTH
AAA
I;CC
I
DDD
FIGURE 3-1
Three Dimensional Schedule Space
-34-
As with any form of mechanical or architectural
drawing the solution to
this dilemma lies in somehow eliminating one of the dimensions so that a two
dimensional drawing can be made.
In doing this it is important to preserve as
much of the information contained in the original representation
as is possible.
Often combinations of different views or cross sections is necessary to completely
represent the original information.
The following discussion describes
sections used separately
the possible projections
and cross
or in combination to compress the three dimensional
picture to two and then to one dimension. Many of the resulting representations
are used by schedulers today and have been named.
Some variations exist for
many of those basic drawings. Still, all customizations are members of one of the
basic families of compressed schedule space drawings.
3.1.1
Two Dimensional Projections
Three projections, one parallel to each of the three axis, are possible.
However, the two projections onto the time-east/west and the time-north/south,
plane are very similar.
These will be referred to jointly as schedule maps. The
other projection onto the east/west - nortVsouth plane is a quite different
representation
known as a route map.
Route Map
Route maps (see Figure 3-2a) are probably the most familiar representation
of an airline network. They actually depict the location of each station and the
links between stations. However, three significant portions of the original
-35-
10:00
13:40
11:20
;3
a.
b. INDEPENDENT FLIGHTS
LINKS
......
f---- ---1\
- ----
-
I
rl)
FREQUENCY
1/2
_ - - _2/2
c. BIDIRECTIONAL
FREQUENCY
FIGURE 3-2
Route Maps
- -
-2ii/
~~~3,!l
- --
d. MULTIPLE ROUTING PATTERNS
-36-
information content are lost in this projection.
time information.
Most important is the loss of all
The essence of a schedule over a mere route plan is completely
lost. Turns and crossovers cannot be shown-this
loss is critical.
Almost as significant is the loss of identity of individual routings.
All
segments between two stations overlap so that it is impossible to determine the
frequency of nonstop service in any market.
its name, does not show routings.
Furthermore,
a route map, despite
Only links are depicted on this map; the
chronological sequence of segments and flights is lost:
through service is not
depicted, only nonstops can be inferred.
These last two deficiencies are partially a result of the loss of the time
variable since frequency and chronological sequence are time dependent relationships. But it is also true that all three of these losses result from the overlapping
of events that are independent when viewed in space-time but are identical when
only their spatial dimensions are considered.
For instance, if only one flight were
shown on a route map each of these problems would be resolved.
In Figure 3-2b two flights that have no common segments are shown and
labeled with operating times.
Here it is possible to see most of the schedule's
dimensions including the frequency (one) and the routings.
Still, without a time
axis it is difficult to visualize the relationships between many events occuring
throughout the system simultaneously. In addition, it is difficult to indicate which
of the potential service offerings have been excluded.
To alleviate the overlapping problem to some extent, a number of methods
can be used. Each link on a route map can be labeled with bidirectional frequency
-37of flight segments. In Figure 3-2c the frequency from the first (alphabetically)
station is shown before the reverse direction.
This provides nonstop service
information but still does nothing to depict through service or service times.
The use of color, or other discriminating drawing technique such as the
broken lines used in Figure 3-2d, makes it possible to indicate which segments are
members of the same routing pattern.
In this representation
not only are nonstop
frequencies shown but all through services are shown as well.
The crisscrossing
lines within the station symbols indicate how flights turn to one another. If drawn
large enough flight numbers and operating times could be added to this map and a
complete representation of the original information could be achieved.
However, it is easy to see that if many different routings were specified
over the same set of stations it would become increasingly difficult to find unique
colors or line dotting patterns to represent each routing. In high density markets
the number of lines that will have to be drawn will become cumbersome.
This
problem will aggrevate the confusion of links that cross each other or that must
be drawn as arcs to avoid passing through intervening stations.
Short haul, high
density regions become tremendously cluttered with the volume of service needed
to serve them. These problems, combined with the difficulty of poor depiction of
time and time dependent relationships such as crossovers, make route maps poor
candidates for primary schedule development tools.
Route maps do have some valuable functions, however.
They provide a
quick and easy way to grasp an overview of the network geography and link
patterns.
They can also be used to display numerical data about each individual
link or market.
These provide valuable reference
data to schedulers during
-38-
schedule development.
In addition to frequency, route maps can be labeled with
costs, fares, origin-destination traffic, on board traffic, distances or block times.
Schedule Map
of the
Schedule maps (see Figure 3-3) are the most basic representation
schedule puzzle.
They preserve all of the important information about the
schedule while simplifying the representation
tremendously.
They are often used
by schedulers for small networks; however, as we will see, they become cumbersome for larger networks.
The projection of a three dimensional schedule space must be perpendicular
to the time axis, however, it may be done parallel to either the north/south or
east/west
axis.
In fact, any intermediate
axis would also be satisfactory.
This
arbitrary axis will be labeled the S (for space) axis.
However, an immediate
recognized.
problem with any of these projections can be
Stations will be located on the S-axis with varying spacing.
Some
stations may even overlap or be excessively far apart thereby wasting space on
what will likely be a very large chart already.
This problem is usually solved by mapping, rather than simply projecting
each station onto the S-axis stations are placed at equal distances along the Saxis.
The width of each station is varied to permit the necessary number of
sequence lines to pass through them without overlapping.
This width,
in fact,
corresponds directly to the number of gates available at each station (see below).
Stations may be mapped in geographical order from west to east or
-39EST
7:00
8:00
9:00
10:00
w
2
11:00
12:00
13:00
14:00
15:00
-40north to south, or they may be rearranged in any convenient order. For example,
if one routing is flown frequently, the stations in this routing may be shown
adjacent to one another. The only information lost here is the actual distance
between stations and the direction of each segment. Distance information can be
inferred from the block time of a segment. Directional information, meaning
whether the aircraft
is flying north, south, east or west, is of interest
in
determining the circuity of a flight, but it is not difficult to refer back to a route
map for this information.
The sequence lines are often labeled with additional data to enhance the
value of the chart.
Arrival and departure time may be shown next to these
operations since reading directly from the time axis is somewhat inaccurate.
When this is done the hour is usually dropped and only the minutes past the hour
shown, since the time axis is sufficient to infer the hour. Flight numbers, traffic,
cost or other information can be indicated along each segment line.
The value of a schedule map is that it depicts the pieces of the schedule
puzzle in both time and space dimensions.
Relationships between flights such as
turns and crossovers are readily seen. Patterns of service can be inferred directly
by noting the number and times of flights linking two stations.
Through service is
also depicted because flight sequence lines are drawn as continuous space-time
lines on the chart.
The problem of overlapping lines is not as severe since two
services in a single market are unlikely to occur at the same time, and stations
can be represented not as single locations but as a set of "gates".
No matter how the stations are arranged, some routings will join stations
that are separated by one or more intervening stations on the s-axis.
The lines
-41connecting these stations must pass over the intervening stations and mingle with
the nonstop segment lines in these markets.
Flights that actually proceed on a
very straight course providing good through service between two points, may be
depicted by long lines that not only cross over many other stations, but may even
reverse direction on the schedule map.
The problem becomes quite severe when
many sequences are shown simultaneously.
The web of lines becomes so tangled
that these charts are often referred to as "spider charts".
Sequences become
difficult to trace, service patterns in various markets become obscured and room
for auxiliary information such as precise times and flight numbers is wiped out.
Congestion is particularly acute at connecting banks. The schedule map shows so
much information at once that it can't all be drawn in one place.
Both two dimensional representations
problem.
of a schedule suffer from this
Too much information is compacted on one piece of paper resulting in
confusion and loss of vital information.
untangling this information.
The solution lies in extracting
or
Four ways exist to extract one dimensional subsets
of the schedule map that are often far more convenient to use.
3.1.2
Cross-Section
Views of Schedule Maps
Schedule maps can be cut or cross-sectioned
the s-axis.
A cross-section
along either the time axis or
at a single point in time provides very little visual
information but does serve to count aircraft.
Cross-sections of the s-axis can be
made either at a single station, producing a station activity or ramp chart, or
between stations, producing a market profile.
-42Aircraft Count
Sequences must be continuous descriptions of the activity
aircraft.
of a single
All flights arriving at a station either turn to subsequent flights or
remain at that station until the end of the schedule period. Similarly, all flights
departing a station must turn from a prior flight or must have been located at
that station since the beginning of the period.
When the schedule map is cut at a single time across the entire network,
all sequences will be found to be either enroute or on the ground at one of the
stations.
The number of sequence lines crossing any time cross section must be
identical for any time chosen.
This sum is the number of aircraft needed to fly
this schedule.
This is, of course, a valuable procedure,
but very little information is
retained by this cross section of limited value in the entire schedule development
process.
Events may occur continuously through time and thus an infinite number
of those cross sections would theoretically be needed to represent the entire
schedule.
Station and Market Profiles
In contrast to the continuity of the time dimension, the space dimension
consists of a finite number of discrete stations and markets.
sections along these discrete
development tools.
spacial elements
This makes cross
far more valuable as schedule
-43Figure 3-4a shows a basic station activity chart.
Each event, arrival or
departure, is labeled with the number of the flight involved. When charts for each
station are used together the complete schedule can be represented.
Many of the difficulties
representation.
of the schedule
map are eliminated
The confusion of criss-crossing lines is eliminated.
in this
The routings
of individual flights may still be traced through the flight number labels, although
this is clearly more cumbersome.
The labels on each arrival and departure can
include the upline and downline itineraries,
respectively.
This addition makes the
tracing of flights through the system feasible.
Overlapping still occurs in both the time and space dimensions on the
station activity chart. Many flights may arrive or depart at nearly the same time,
making the chart difficult to draw. To solve this problem, clusters of arrivals and
departures can be mapped to equidistant locations on the time axis and then
actual operation times can be shown. In the space dimension, because it is only a
single point, all events occuring simultaneously at that location overlap. This
problem can be alleviated by expanding the cross section from a single station to
a collection of gates.
Each flight sequence is assigned to a gate and remains
there until its departure.
No two flights can occupy a gate location simul-
taneously, thereby eliminating any overlap.
Figure 3-4b shows a fully labeled station activity chart with gates.
It is
evident that this is a much more valuable tool than the simple version in Figure 3-
4a.
In this chart it is possible to trace not only flight routings thorugh the
itinerary tables, but also to see turns occuring at the station. Possible crossovers
-44AAA
a.
n
0.UU
r
101
t-
BBB CCC 1(
--
102 -
0
XXX
L
10:00 .
-o
r
103
1
101 DDD
102
102 NNN
1(
,--*.---o.
11:00
YYY
101
104
105
106
04 EEE FFF
DDD CCC 1(
XXX YYY 1(
-I
F106
107 NNN
105 PPP
105
12:00
b.
a. BASIC STATION ACTI\ TY
At'
A
:U
EXPANDED STATION ACTIVITY
'T
n nn
Z.Uu
.n
LU-
.O
3
4
p
0
:20
0
0
:35 -
:35
:55 -
C
10:00
:55
o
.20
:20
I
:40
11:00
:00
:00
:20
:451
12:00
1;_
c
:55-
c. RAMP
:45
_
:55
CHART
d. ACTIVITY PROFILE
FIGURE 3-4
Station Activity Representations
-45-
can be readily found on this diagram by looking for flights with compatible
operating times.
It should be noted that these charts are not very efficient in terms of the
area needed to represent them.
Itineraries and flight numbers must be repeated
at each station that a flight uses, and the turn information is represented by a
multitude of lines at each station which would be avoided if flight itineraries were
shown strung together on a single line. This concept will be discussed in the next
section.
Another version of the station profile is shown in Figure 3-4c.
known as a ramp chart.
representation
Upline and downline itineraries
of gates is retained.
This is
are dropped, but the
Flight numbers still provide a means of
tracing the flight's itinerary. In addition, the times needed for maneuvering to
the gate and for the deplaning, enplaning, and cabin cleaning activities are each
indicated separately from any slack, or unused time on the aircraft.
Ramp charts are useful in isolating the problems of gate and station charts
are useful in isolating the problems of gate and station facilities
Periods of peak activity
can be spotted readily.
utilization.
In addition, maintenance
scheduling can be done using this chart to determine when and if a piece of
equipment is at a maintenance base for a sufficient period to permit necessary
maintenance to be performed.
Another important application of the ramp chart is for determining
coverage.
The slack periods on each of the aircraft can be integrated to produce
a discrete function over time (see Figure 3-4d) of the number of free aircraft on
-46the ramp known as a station activity
incremental
count.
profile.
This function known as the
The function is increased by one for each arrival and
decreased for each departure of the same equipment type. This function reveals a
number of interesting things about the schedule. First of all, while the function is
non-zero, at least one aircraft
is always sitting on the ground unused.
This
situation may be planned so that one aircraft will always be available to take over
for, or "cover", a malfunctioning one. When the function stays above any number
for an extended period of time, this many aircraft are always available. It may be
desirable to use these extra aircraft.
equipment is actually
Note that, depending on how the physical
turned, these "aircraft"
individual pieces of equipment.
may be composed of different
By returning some flights the actual aircraft can
be freed during this period.
When an aircraft is available for an extended continuous period except for
a short interruption, this will be seen as a small trough in the function.
A minor
change in scheduled flight times could close this gap and free an aircraft
for
additional flights or reduce the minimum number of aircraft of this type required
to fly the schedule.
-47-
3.1.3
Projections of Schedule Maps
Probably the most powerful representations
of a schedule are projections
of the schedule map onto either of the two remaining axes.
A schedule map
projected onto the time axis is termed a sequence chart here, although the
misleading term "cycle chart" is more commonly used.
When projected onto the
s-axis, the resulting chart is referred to as a routing chart.
It should be noted
that the term "routing chart" is used by some schedulers to refer to station
activity
and sequence charts as well, but for clarity, only the more specific
definition.
Both sequence and routing charts are not actually true projections.
Rather, each sequence is projected individually into a one-dimensional form.
These charts consist of a collection of one dimensional representations.
The
salient feature of this technique is that it untangles the confused web of lines
found on most of the preceding charts.
These charts are thus far more readable
and many important relationships are easily discerned.
Routing Chart
A routing chart results by projecting portions of flight sequences onto the
s-axis.
In order for these projections to be one dimensional, the portions
projected must be unidirectional.
For example, a sequence that proceeds from
left to right on the schedule map and then reverses direction is shown in Figure 35a.
When projected,
this sequence produces two lines of the routing chart
connected by an arc to represent the turn.
elements.
These lines will be called sequence
Each sequence element normally occupies one line on the routing
chart. Space can be saved by using lines for more than one of these sequence
-48AAA
DDD
.
_CCC
BBB
EEE
l
8:50
8:50
8 :00
9:15
,
.
.
10:00
101
_
13:00
11:40
_102 12:15
. _
.
p
.
_
.
,
10:30
)
f
L
21:30
7:00
.
20:00 19:20
104
-q
14:30
.
13
10
~~~3
,
_
.
f
13:45
_L
lw
15:00
.
16:00
J.,
l
l
_
_
.
.
_
11:40
13:30 12:50
105
__ ,
11:10
15:10
.
l_
p
17:50
18:15
10:00
9:1 5
102_
.
;
I
_
)
a. BASIC ROUTING CHART
AAA
c(-
I
(,
F
BBB
11:45
101
10:30 10:00
9:00
12:30
102
13:40 14:30
15:30
13:45
107
12:30 12:00
11:00
14:30
108
15:40 16:20
17:30
19:15
(
103
18:30 17:50
104
21:00
7:45
105
7:00
!8:30
rf
106
.
1
i
l
1
EEE
19:10 19:50 ~-
18:00
...i,
108
.
--
21:00
9:00
110
9:40
111
10:30
I---
-
._
_.
._
£
r
I
13:00 112 14:1U0
16:30
Hi
L
16:10 113 15:00
&
17:00
18;10 19:00
21:15 20:40
22:30
-
- -
1015 i
i
b. REARRANGED ROUTING CHART
Routing Charts
_
109
1 2:00
FIG URE 3-5
20:30 I
I 21:40
c1
I
.
20:15
DDD
m
,
; ;c,
9:1
I O
3"
r
CCC
-
w
114
19:40
115
20:00
*
l-
-49portions provided they use mutually exclusive sets of stations and thus do not
overlap.
Operating times must be displayed in numeric form since the time axis has
been eliminated.
Because time is no longer a factor in the order, sequence
elements can be arranged in any order on the chart.
Often, successive sequence
elements belonging to the same sequence are shown adjacent to one another. This
eliminates all crossing lines. It may, however, be more convenient to show those
sequence elements that follow identical or similar routings together.
In this form
(see Figure 3-5b) patterns of service in individual markets are easily visualized.
Sequences shown thus may well cross over one another producing some confusion,
but this may be warranted by the increased clarity of the isolated service
patterns.
Although routing charts do untangle the schedule representation, separate
service patterns and depict complete sequences including turns, this diagram has
serious drawbacks.
Time information is not graphically depicted and thus many
important relationships do not stand out. Turns, crossovers, gate utilization and
coverage information is not clearly depicted. The importance of these situations
and other space-time relationships makes routing charts relatively poor schedule
development tools.
They are, however, useful for presenting finished schedules
because of their clear and compact representation of service patterns.
-50-
Sequence Chart
Sequence charts combine the three most important features found in the
other charts. First of all, because sequences are projected onto the time axis, the
important time relationships are maintained. Secondly, individual sequences are
untangled into separate lines, each representing the flow of a single aircraft.
No
retracing or crisscrossing is necessary. Finally, turns are clearly depicted without
explicit turn lines that point from one flight to the next. Each flight is physically
located adjacent to its logical predecessor.
Figure 3-6 shows a typical sequence chart.
(Because of the confusion
between the terms "sequence" and "cycle" this is often called a "cycle chart.")
In
this chart each stop is shown in its proper location along the time axis and is
labeled with the minutes portion of the time to help pinpoint each event.
Each
line, representing one sequence, is labeled with an aircraft type.
Sequence charts do not permit one to view the operations
station easily to find turns or crossovers.
at a single
They do, however, provide an easily
interpreted and easily drawn display of the information contained in the schedule.
Their other main drawback is the fact that to move a flight from one sequence to
another requires that it be completely erased and redrawn.
When done manually,
this involves a great deal of effort for a relatively simple modification.
Once a number of stops at a station has been identified, it is easy to see
whether they line up properly to permit turns or crossovers. The effect of sliding
a flight can also be easily discerned.
The flight can be thought of as physical
markers on a track that can be pushed back and forth as appropriate points are
aligned. The spacing of services throughout the day can be observed to determine
-51-
FqT
C-V
DC9
10:00
11:00
10
15
I I I' .
I i
JJJ
17
30
I
IoDD
· .
r,
r- J&_
ULJL
:U
I
1
15'
--
I
I
r
I
DD
D
-I -
55
LLL
50
45L
I
TTT
.
I
I
I
25
.
L
40
50
-EEE
20
00
10
-
r
I:f/ -L
I&
KK
I I/
20
Lr L
~
uze
rrt '
15
45
I
00
I
40
I
05
SSS
I!
I
I
`"""'
I
vv¥vv
"?'
,
VV
45
~
Ii
30,
IAI
I
10
45
I KKK 1
CCC
-I
00
40
I yyy -
_
50
20
-----I35 TTT
L~ I
FI
V
_
I 1\\ \ r- - -- --
-J
I
t
-4 r xlvlt ,J
50
25
15
[E7
r7
--
30
05
DC9
.
15:00
I
-
10
727
DC9
I
c
30
,JJJ i
727
DC9
14:00
I
00
-
15
50r
-
10
I
I
40
40
13:00
50
:-.
I
.,
IAAA
YYY
IAAI
DC9
727
--r I
I
L
I
DC9
12:00
20
1.......
T T T
00
30
-
I
50
I EF
I
I
_
!
FIGURE 3-6
Typical Sequence Chart
15
i
-52-
that appropriate coverage of each market has been provided. Stops at each
station, once found, can be evaluated to determine if a congestion problem will
occur. Alternative turns can be easily evaluated by observing the turns at each
station to determine whether a flight could fit into another sequence without
being slid.
In general, the sequence chart is the best overall display for
performing all of the necessary scheduling functions.
3.2
Prior Applications of Computers for Schedule Development
In the past, the complex, difficult and important task of scheduling has
been studied by schedulers, computer analysts and operations researchers.
Many
computer programs have been developed to aid schedulers or even take over some
of their functions.
However, as far as is known to this author, no airline has
considered replacing its scheduling department with a computer or even including
a computer program as the focal point of the scheduling process. Of course there
are political reasons why this may never happen; even so, much of the failure of
these projects can be attributed to the limitations of the application programs
themselves.
Existing programs can be categorized into three groups. The first includes
the attempts to generate desirable and valid schedules through automated
algorithms. The second include batch processing applications developed to enter,
store, validate, format and distribute schedule information.
Finally, we have seen
the recent emergence of online programs to permit entry, manipulation and
analysis of schedules as well as access to pertinent traffic, cost and revenue data.
Due to inadequate
command languages, manipulative
capability
and graphics
-53-
support as well as limiting assumptions, these programs are inadequate sucessors
to current manual schedule development procedures.
3.2.1
Automated Programs
Numerous programs have been developed to optimize
or produce good
heuristic solutions to various segments of the scheduling problem.
Separate
problems are usually defined for determining the routing of flights, the optimal
frequency of these routings, the dispatching of these flights at optimal times of
the day, and the turning of aircraft to minimize fleet size while meeting
maintenance and station constraints.
Simpson (1969) describes how linear programming, integer linear programming and combinatorial programming techniques can be used to solve many of
these problems. However, these techniques are limited by the assumptions and
subjective input data that they require. Frequency/aircraft type determination,
known as fleet assignment, requires an estimation of the functional relationship
between frequency and demand and dispatching models use a curve of demand by
time of day in their objective functions.
Cost and revenue estimations are also
required for some of these procedures.
Such explicit expression of subjective
information is not only difficult and time consuming, but inaccuracies will lead to
nonoptimal solutions.
Another problem with these techniques is that they only consider a limited
number of the factors that influence the development of a good schedule.
Historical service patterns often need to be preserved,
fit into the optimized schedule.
even though they do not
Some routings, schedule times and turns may be
-54-
considered important by the scheduler for marketing or operational reasons that
cannot be expressed in the optimization program.
A major drawback of these solution techniques
is their
independent
approach to each segment of the overall problem. No feedback mechanisms exist
from one procedure to another that permits reformulation of one problem in light
of the outcome of a latter one. In addition, some of the algorithms themselves
only provide local optima. For instance, the dispatching models optimize for each
route but do not consider the interdependent relationships between flights needed
to form crossovers.
In fact, crossovers are ignored or only considered as an
afterthought by all of these algorithms.
Inspite of these problems, some algorithms play an important role as a
supplement to the scheduling process.
They often provide a first cut in situations
such as mergers or in new markets where past experience can't be used as a guide.
Aircraft routing models are particularly useful since they are the least subjective.
A variety of systems designed to schedule aircraft
number of foreign and domestic air carriers.*
have been implemented by a
While none of these has replaced
schedulers, they all provide valuable tools for designing and evaluating schedules.
3.2.2
Batch Processing Aids
To deal with the complexity and detail of a schedule with sufficient
accuracy and speed, a number of tools have been developed to translate schedule
*See:
Etschmair; Frey; Girard; Labomba:da;
Useros; Walker-Powell and Williamson.
Loughran; Mathaisel; Niederer;
-55-
data from manual media to machine readable form.
Validation, summarization,
evaluation and vizualization of the schedule can be performed once it has been
entered.
These functions aid the scheduler in the iterative process of schedule
development even though they have no immediate input into the initial development of the schedule.
These programs are usually designed to accept completed schedules in the
form of punched cards with information on each flight. Once entered, the number
of aircraft used, utilization of aircraft and station facilities, average segment
length and other global parameters can be computed.
Checks can be made for
balanced flow, sufficient maintenance time, accuracy of manually computed block
times, flight number repetition, connection feasibility, and many other details
that may have been overlooked or erroneously specified.
Reports can also be generated by these programs that allow rapid, accurate
and concise communication of a proposed schedule to various departments within
the airline. Reports on station activity, maintenance time allotments and service
patterns can easily be produced allowing each department to see that information
which is important to it in the format it is best able to use. Maintaining schedules
within a computer facilitates
long term storage and permits quick retrieval of
historical schedule information.
The availability of proposed schedules in computer readable form also
permits operations researchers to analyze the schedule using previously prepared
programs.
Although they cannot be used in the original schedule development,
these analyses may provide important
feedback to schedulers on how certain
-56-
aspects of the schedule might be improved or how profitable
the proposed
schedule is likely to be.
3.2.3
Online Scheduling Aids
The final category
of scheduling tools are the most powerful and the
closest approximation to the proposed schedule development systems.
they are online, they can provide schedulers with instantaneous
questions and validation of input data.
Because
responses to
These systems are the true forerunners of
the current work in that they recognized the essential role of human judgment in
the scheduling process and attempt to aid the scheduler by providing tools.
Three such systems are known to this author.
One developed by Potomac
Scheduling, Inc., called SPS provides online access to an extensive service, traffic
and operating statistics data base as well as models for forecasting the impact of
schedule modifications.
first known attempt
The second system, developed at Eastern Airlines, is the
to provide a tool for maintaining a schedule data base
concurrently with the manually produced form. The third system, developed at
the Flight Transportation Laboratory at M.I.T., known as the Competitive Airline
Strategy Simulation (CASS) contains a powerful scheduling tool known as the PFP
as well as traffic allocation and operational forecasting capability.
Potomac Scheduling's SPS (Labovitz,1978) was designed for and with the
help of American Airlines.
As a result, it permits all of the flexibility to handle
complex schedules with varying days of application, multiple schedule periods and
-57-
multiple versions of a single schedule proposal. Without these features, a system
could only have theoretical or academic but not practical commercial value.
SPS performs two major functions.
It provides an extremely
flexible
command language for accessing the existing data base, which includes complete
current and historical schedules and statistical information on traffic, costs and
revenue.
Its second function is to accept proposed schedules and forecast their
value. To do this, some commands are provided to enter and manipulate flights.
Other commands call back flight information for review and modification.
A
scheduler could presumably sit at a terminal and design a schedule with no other
assistance.
However, this is not a realistic expectation given the limitations of SPS. It
was designed primarily to analyze schedule changes in terms of their marketing
and operational impact rather than to act as the primary medium for develop-
ment. For instance, flights are specified without regard to turns, which is not
important to the analysis of flights but is to the operation of the schedule.
Displays are all in the form of tabular reports which are produced on teletype
equipment. No facility exists for sliding flights and crossovers are not expressed
explicitly; so no warning is produced when a deletion or modification makes one
infeasible.
These deficiencies make SPS a poor substitute for manual methods,
although it is a valuable reference tool that can provide direct assistance to
schedulers.
Eastern Airline's system (Smith and Kyle, 1972) while not providing access
to competitive
interactive
schedules or forecasting
capability, comes
closer to being an
graphics tool that a scheduler might use to replace wall charts and
-58-
other manual techniques.
In fact, this is the expressed purpose of the project,
although subsequent inquiry indicates that it was unsuccessful in reaching this
goal.
This system uses three CRTs,
work station for one operator to use.
a line printer,
and a keyboard arranged in a
The CRTs simultaneously display data on
flights, summarizations and statistics for the entire schedule and diagnostics that
aid the user in finding errors.
Like SPS, this system has the complexity necessary to represent actual
commercial
schedules.
neglects crossovers.
It also handles information
on turns although it too
Its use has been limited to translation
of wall charts to
electronic storage, a task which is performed by a keyboard operator,
scheduler.
not a
Its primary failure appears to be that only individual flights can be
shown; relationships between flights or groups of flights relevant to a particular
decision are not displayed together in an easily comprehended format.
M.I.T's CASS system
traffic,
resembles SPS in that it provides the user with
revenue and cost estimations for hypothetical schedules.
However, it is
the schedule development tool included in CASS, known as the PFP, that is of
interest here. The PFP differs profoundly from the SPS counterpart.
The PFP (Elias, November, 1979) is much more of a theoretical foundation
for an advanced scheduling tool than a practical application for commercial use.
Only daily schedules can be handled, turns are not included in the data base and
multiple itinerary flights can't be handled. Crossovers are not specified but
/
-59individual connecting services can be and the validity of these is checked against
global crossover parameters.
Even though the PFP uses a line printer for input and output, it provides a
very interactive environment with immediate feedback on errors, market and
station summaries, balanced flow, equipment utilization and other useful information. Simple sliding operations can be performed as well as other modifications
of basic flight data. Station activity profiles can be produced to assist in turn
decisions even though these decisions must be recorded externally to the program.
In spite of its many valuable features and explicit design as a schedule development tool, the PFP doesn't go far enough to be considered a practical scheduling
tool.
-60-
(This page intentionally left blank.)
-61-
4.
INTERACTIVE GRAPHICS DESIGN FOR SCHEDULE DEVELOPMENT
4.1
The Advantages of Interactive Graphics
No computerized tool yet designed has been accepted by schedulers to assit
them in the development.process. This is certainly the fault of the tools that
have been offered to date.
None of them are capable of becoming an integral
part of the scheduling process.
They perform certain useful functions such as
validity confirmation and report production to facilitate
communications;
some
even act as a desk top calculator to perfom some routine calculations for the
scheduler.
For a computer system to play a central role it must meet two essential
requirements.
First, it must be able to accept and process a scheduler's decisions
in the time frame in which they are made.
For it to provide any support to a
scheduler in his decision making process it must be involved in these decisions as
they are made. Responses to any decision must be made immediately if they are
to be of any value to a scheduler.
It does very little good to have information
about a problem that has already been resolved.
The second requirement for a computer tool designed to be part of the
scheduling process is that it must become the primary medium on which the
schedule is developed, stored and displayed.
manually drawn wall charts.
Lo.ic.iein
orler
In other words it has to replace the
If a scheduler feels he must develop the schedule
io grasp its complexities, he is unlikely to take the time to
ria-s.-ribe the information from manual to electronic form.
This task would be
-62easily to understand and use.
Without graphics and interactive
computer is worse than useless to a scheduler-it
But why should a scheduler
capability a
is a nuisance.
use a computer
rather
than the familiar
techniques he has always used if it merely replicates the manual methods?
The
capabilities and advantages of using a computer to develop schedules goes far
beyond its ability to merely replacate
manual methods.
If fully utilized, the
computer can dramatically alter the schedule development process. Many aspects
of the development can be simplified and the inherent complexities can be more
easily visualized and understood. The speed at which schedules are developed can
be increased substantially,
making more intricate and efficient schedules possible
while simultaneously increasing the number of schedule alternatives
airline
can choose from.
Finally, communications
between
that the
the scheduling
function and other parts of the organization can be improved in quality and in
speed.
4.1.1
Simplification
Many routine
tasks that
a scheduler
must now do manually can be
performed more efficiently by a computer. These tasks are those that can be
fully specified in advance and which are performed repeatedly.
Thus, a scheduler
-63-
Thus, a scheduler can be relieved of the tedious and mindless work that presently
bogs down his decision making ability throughout development.
A
number of cc rnrLutationsand data validations must be accomplished.
Block times are computed, time constraints are checked for consistency and over
constraint, aircraft types and stations are checked for validity, operating
restrictions are checked, stage lengths and circuity are computed, costs and fuel
use ae estilnated and flight numbers are cross checked for repetition.
As the information for a flight is entered all appropriate
computations will be made automatically.
checks and
In this way an error can be detected
before continuing the entry of an already invalid or inappropriate flight. This can
save a great deal of time that would have to be spent retracting
later to find an error.
the information
Before a flight is entered as complete, a check is made to
determine that ail required data is present.
As we have seen, scheduling involves a great deal of complexity in terms of
the number of constraints, opportunities and relationships that must be considered
simultaneously.
User specified constraints,
turns and crossovers must all be
consietered e-ach timrnea small modification is made.
Many simple tasks involve
tedious bookkeeping that not only wastes a scheduler's time but is prone to errors.
Calculations of such things as average utilization broken down by daily sequence
and aggregated by period, aircraft type or airline are exceedingly time consuming.
Because of this, a human may only perform these tasks infrequently after large
portions of a schedule have been completed.
-64-
While a computer can not produce a valid schedule, it can validate one
against a predetermined
set of criteria
and help a schedule find errors
and
opportunities to correct them. The computer can count aircraft using a simple
preset algorithm.
It can find discontinuous flow in a sequence or sequences that
simply stop and do not turn to other flights.
Maintenance requirements can be
specified and each sequence checked for compliance.
A computer can assist by
helping the scheduler trace the consequences of hypothetical modifications. By
relieving the scheduler of the tedious burden of manually tracing long chain
reactions through a complex schedule he will be able to consider many more
possibilities and find solutions that were previously too complex to find.
Even
though the computer makes no decisions for him, a computer can make information available to a scheduler during the development process when it is of more
use and may influence his decisions.
This is particularly
important for station
activity and market evaluations that can directly affect a scheduling decision.
4.1.2
Visualization
As earlier
representations.
discussions have indicated,
scheduling lends itself to visual
However, as also noted, no single format adequately captures all
of the complexities
or the quantity of information
required. Before making a
decision all pertinent information must first be found -- a difficult task in itself.
A visual search combined with some memorized or written indexes must be used
to scan the room full of wall charts to find all flights relating to one problem.
Even then, the depiction of all this information will remain in disjoint locations
making it difficult to get a good visual grasp of the problem.
It is often necessary
to view a schedule through multiple projections or vantage points.
It is also
convenient to be able to extract and untangle only that portion of the schedule
-65-
that relates to the decision at hand. Changing one flight on only one wall chart
may involve tedious erasing and redrawing. On more than one chart, changes may
be so difficult to keep up with that schedulers are forced to limit themselves to
one general chart.
Beyond its ability to emulate the manual graphics tools-wall charts-a
computer can easily translate
data from a format convenient for one task or
phase of the development process, to another format.
It can also bring all of the
data needed for a single decision together in one display on an ad-hoc basis. While
this adds no new or derived information to the problem, it does vastly improve the
scheduler's ability to use the existing information fully. Furthermore, the computer finds and collects this information faster with far less physical and mental
effort on the part of the scheduler. Finally, modifications can be effected with
minimal physical effort on the part of the scheduler.
4.1.3
Communication
Including a computer in the scheduling process has important ramifications
beyond its ability to aid an individual scheduler with his own job. By placing the
schedule in a computer where it can be accessed repeatedly by multiple users
throughout the development period, communication, both among schedulers and
between schedulers and other concerned groups within the airline, is enhanced
dramatically.
Improved communications results not only in better schedules but
in better airline planning as a whole.
Schedulers already ha/e standardized ways of describing and representing
the information they need to use and communicate.
However, it is not always
-66-
easy to recreate and describe the thought process that leads up to a decision or
the discovery of a problem.
It is often a complicated series of discoveries and
insights that bring a scheduler to a particular
decision or the realization that a
conflict exists in the schedule.
On a wall chart one can not draw this thought process as it leads from one
end of the room to the other. But by reproducing the successive displays that one
scheduler
his thoughts and relate
them to another
A problem that one scheduler found on a wall chart
may be difficult
used, he can reconstruct
scheduler.
to find again, while a problem found by the computer or that clearly stands out on
a particular display can easily be recalled and dealt with.
To the extent that any differences do exist between schedulers in the way
they construct
standardize
and represent
them.
schedule information,
This alone will facilitate
the computer
communication.
will help
It will help new
employees learn procedures faster and provide a common framework for schedule
developers working on current and future schedules.
More importantly, communication between schedulers and the rest of the
organization
can be improved.
Programs
already exist to help a clerk to
transcribe a schedule from wall charts to machine readable form. These programs
then can produce reports for operations, upper management, traffic forecasting
and other interested departments.
Alone, this capability speeds the process of
diseminating the schedule so that it may be reviewed, problems found, suggestions
made, and operational
plans prepared.
The iterative
schedule refinements can thus be accomplished.
process of successive
-67-
With the computer involved directly in the schedule development process,
communication can be continuous.
schedule,
review
becomes
Rather than just reviewing the completed
an ongoing process.
This involves
the
other
management departments in some decisions and allows appropriate modifications
to be made before the schedule is locked in and difficult to modify. This ongoing
communication
with the scheduling function
makes other
departments
more
cognizant of the problems schedulers face and enhances the responsiveness of
schedulers to other parts of the organization.
Computer readable schedules open up another door that holds considerable
promise.
It will allow more subjective computer analysis incorporating modern
operations research techniques to be performed. For instance, traffic allocation
and revenue forecasting programs such as CASS can be applied to schedules as
they are developed to aid the scheduler. While this kind of analysis is useful after
a schedule is completed, it does not to have the impact it will have when it can be
applied during the development.
When the scheduler has immediate feedback on
the impact of a decision this information will become much more valuable and
have a larger impact on the schedule.
Finally, communications will improve simply by speeding up the scheduler's
job, allowing more schedules to be produced in a given period of time.
While only
one iteration of a schedule is now often the maximum possible, computer aided
scheduling
mkaes many iterations
feasible.
This permits
far more of the
suggestions of other departments to be evaluated before the schedule is implemented; consequently problems can be detected in the planning phase rather than
after they arise during actual operations.
-684.2
The Advantages of Interactive Graphics
As we have seen, the schedule development process is very complex and
requires
evaluated.
that an intricate
schedule be produced, manipulated,
validated
and
Each phase of this process has somewhat different requirements in
terms of both the kind of information that must be available and the way in which
it is displayed.
Before going on to discuss the advantages
and design of
interactive graphics to aid the scheduler, each of the identifiable phases of the
scheduling process will be evaluated to determine the differences in information
requirements for each phase.
It should be noted that these phases are not chronologically distinct, rather
they are unique in terms of the activities
and information
required by each.
Although portions of a schedule may be produced, manipulated, validated and
evaluated,
in that order, it is frequently necessary to revert to an earlier phase.
After an initial validation uncovers a deficiency, additional pieces of the schedule
must be produced and/or manipulated.
In fact, the validation of a particular
addition or modification may be made immediately, and thus these phases may in
effect occur simultaneously.
4.2.1
Production
Schedule production is the function of constructing flights.
A flight in its
most basic form consists of a unique flight number, an aircraft type, a number of
routings, validity dates for each routing such that no more than one routing is
valid on any given day, and a consistent set of arrival, departure, ground and
flight times.
This is all that is required to define a flight; however, some
-69additional information can be associated with each flight that will be useful in the
manipulation and evaluation phases or will provide valuable reference data for the
scheduler.
The most important addition to the basic information is a set of constraints
on the times set for each itinerary.
Each time required for a unique and
unambiguous definition of an itinerary can be constrained in two ways; either an
absolute or a relative constraint can be set. For instance, an absolute constraint
would be: the departure from station "A" must occur between 7:00 and 7:10 or,
the ground time at station "B" must be between 10 and 20 minutes.
A relative
constraint would be: the arrival at "A" can move no more than 10 minutes earlier,
relative to its nominal value; or, the ground time at B cannot be more than 20
minutes longer than it already is.
These constraints can pin, bracket or limit a departure or arrival time. A
pinned constraint
is one that allows no movement either earlier or later.
A
bracketed time constraint allows movement both earlier or later but only to
certain limits in each direction.
A simple limit constraint only restricts
movement in a single direction.
Time constraints entered by the scheduler with the flight data serve to
help translate subjective discussions into firm, quantifiable values. For instance,
if flight 101 has always departed station A at 9:00 and the continuity of this time
is important, the scheduler has the option of noting this requirement explicitly to
serve as a reminder later in the schedule development process.
As an additional
reminder it may even be valuable to associate a reason for each constraint with
-70-
that constraint for subsequent reference. This is particularly important if more
than one scheduler will be working on a particular schedule.
Other information that a scheduler may wish to associate with a flight
relate to cost and operational requirements. For instance, each segment may
have a marginal operating cost that will be useful during the evaluation phase.
Fuel requirements and fuel loading plans can also be associated with appropriate
segments and stops.
The crew required to fly a particular
included here for tallying later.
leg may also be
Service offerings can be excluded from a flight.
For instance, an itinerary such as A, B, C, D may exclude AD service.
The
inflight services or amenities, such as meals, can be noted here as well.
Many additional pieces of data will be entered
scheduler required.
with a flight as the
The important issue here is that a great deal of nonessential
data may be available that will not be used during certain later phases.
This
requires that the production of a flight be done through different displays and
using different techniques than those used in other phases. The design of tools for
this phase will reflect these requirements.
Manipulation
4.2.2
During the manipulation phase, a scheduler is concerned with the basic
relationships between flights -- crossovers and turns. Using the flight itineraries
produced earlier, the scheduler must be able to vary the operating times of a
flight, (while keeping the rest of the flight's structure), an operation known as
sliding.
-71-
The complex process of turning, crossing and sliding flights can be aided by
appropriate representations
and techniques.
Tools designed for this phase should
help the scheduler grasp both the extensive opportunities available as well as the
numerous restrictions present. Since this phase is only concerned with space-time
relationships, all of the other data associated with each flight itinerary is
necessary. In fact, even the flight number is superfluous, although the time
constraint
information is very valuable here, as are the operating times, the
itineraries and the turn and crossover information.
The fact that individual itineraries
have repeated applications
(days on
which the itinerary is applicable) complicates the process of schedule manipulation and in turn the design of tools to assist it. A particular itinerary may turn or
crossover to a different set of flights on different days. When turning, crossing or
sliding such an itinerary the implications of that action may extend not only to
many flights but to flights on many days.
All of these implications
must be
observed and handled for successful manipulations to occur.
The process that a scheduler uses to go about making modifications makes
it possible to identify those sets of data that will be relevant to any particular
decision.
It can be said that a scheduler makes one decision at a time, thereby
simplifying the problem to supplying information for that decision only. But this
is not, in fact, the case, since the impact of one decision may require another
decision to be made and then another and so on. This is essentially a recursive
process since a whole chain of decisions may be required, each one permitting the
previous one to be made.
This problem is analyzed later; first, the question of
how each individual decision is made independently is considered.
-72If the turning of a single flight can be considered independently then a set
of required information can be identified.
Since a turn occurs at a specific
station and at a specific time it is useful to see all flights that originate and
terminate at that station within a limited band of time. This information supplies
all of the alternative
been designated.
turns that could be made including which ones have already
With this information it is possible to turn one application of a
flight to one application of another flight. Of course, each application of a flight
must be turned separately. Thus, to turn one flight requires that this process be
repeated once for each of the application of each of the flight itineraries.
Unlike a turn, a crossover between two itineraries with multiple applications is valid on every day for which both of those itineraries
apply.
Since
crossovers also occur at a particular station during a narrow timeband, it is again
convenient to produce a set of itineraries that have these characteristics and are
thus candidates for crossovers.
However, in this case, not only will one day's
crossovers be needed, rather all crossovers possible on any of the applicable days
of the itinerary of interest should be included. Each potential crossover must, of
course, be labeled with its applicability.
With this information,
all desirable
crossovers for each application of the flight can be found and specified.
When sliding an itinerary, it is essential to know every relationship that
exists between that flight and other flights. Crossovers for all days of application
must be included as well as every different turn that the itinerary makes. When
the flight is slid, all of the relationships can be observed and any potential
conflicts spotted.
The time constraints placed on the flight when it was first
entered also come into play here.
When a turn, crossover or constraint will be
-73broken by a slide, this information must be made available to the scheduler so
that appropriate modifications can be made.
The issue of recursive decision making indicates
tional set of information.
the need for one addi-
Analysis of a typical recursive decision chain
demonstrates this. Suppose flight 101 must be crossed over to flight 102 but that
flight 102 leaves too early.
Flight 102 must be slid to a later time, but this may
conflict with its turn to flight 103 which departs too soon after 102 to allow 102
to be any later.
later.
To solve this, 102 is turned to another flight, 104, that leaves
Now the flight that turned to 104, 105, is turned to 103 so that the flow
remains continuous. But, conceivably, 105 is too late to turn to 103, so it must be
slid earlier.
This, in turn, could violate a crossover somewhere between 105 and
another flight...and so on; it could be endless.
To help a scheduler find his way through this maze, a summary of each step
can be kept.
This summary would indicate
consequences of that change.
each desired
change and the
When a resolution is finally formulated, the whole
chain can be traversed in reverse effecting each desired change. This process
makes possible review of all the implications of the single initial decision and
evaluation whether it is a worthwhile change.
situations
where a decision contradicts
It is also possible to identify
an earlier
one and is therefore
not
allowable-a deadlock situation. The scheduler would then attempt a different
resolution.
Because of the complexity of the problems encountered in this phase of the
scheduling process many alternative
aid the scheduler.
displays and procedures must be designed to
The primary goal of these tools will be to discover problems
-74and indicate them to the scheduler.
Convenient ways must be found to represent
each of the alternatives and restrictions that the scheduler is faced with, and to
help him find appropriate solutions.
4.2.3
Validation
Each flight, turn, crossover and slide must be individually validated.
the information supplied must meet a certain set of minimum requirements.
All of
For
example, flight numbers can not be duplicated, two itineraries of the same flight
must not have the same application day, turns and crossovers must allow adequate
time between arrivals and departures, etc.
All of this validation can be done on
an individual basis as each entry or decision is made.
However, some validations
are of a more global nature and therefore constitute a separate phase of schedule
development.
Three major forms of global validation
can be discussed.
These are
determinations of aircraft count and of balanced, cyclic and continuous flow in
the network -- aircraft count which were described in Section 2.1.4.
The first
two, however, are only relevant to the boundaries, i.e., the beginning and end of
the schedule period.
Only continuous flow and aircraft
significance to a scheduler.
count are of particular
As long as a consistent and continuous physical flow
of aircraft exists from the beginning to the end of the schedule period, the net
flow of aircraft from one initial state to another final state is only of secondary
importance.
The number of continuous sequences defined by a schedule determine
the number of aircraft that are required to fly it.
-75-
Every flight could be considered to be a sequence by itself.
A single flight
that flies on one day of a schedule period is a continuous sequence for an aircraft
that sits and waits for days, then flys one flight and sits again until the end of the
period. Whenturns are specified between a set of flights, however, these flights
form a longer continuous sequence that can still be flown by one aircraft.
In
tabulating the number of aircraft required to fly a schedule all of the continuous
sequences must be counted, regardless of length.
If two flights are assigned to adjacent positions in a sequence (one before
the other) but are not turned to one another, a discontinuous sequence results.
Until a flight or a turn is added between them that joins the two discontinuous
portions of the sequence, two aircraft are required to fly them.
This indicates a
simple way to count the aircraft requirement of a schedule.
For a set of N
flights, N-1 turns are required to produce a sequence that can be flown by one
aircraft. Note that if the last flight in a sequence is turned to the first, a cycle
is formed, but one aircraft is still required; therefore, cycle-turns are not considered in this calculation.
If the number of non cycle-turns are subtracted from
the number of flights the result will be the number of aircraft required to fly the
schedule as it is currently defined.
It is also useful to count the minimum number of aircraft
required to fly a given schedule.
that would be
This can be compared to the current count to
evaluate the efficiency of the turn pattern.
The minimum is found by performing
an incremental count of aircraft at each station for the entire period.
The
differences between the minimum count and the count at a given time, summed
for all stations and added to the number of aircraft in flight at that time, yields
the minimum aircraft requirement.
-76Three methods of finding inefficiency in the schedule are possible.
The
first two are accomplished by observing the originations and terminations
at a
single station.
If it is found that an aircraft arrives in time to be turned to
another out going flight, but is not yet turned to any flights, the specification of
this turn will eliminate one aircraft from the number required.
The second
method is to observe the pattern of originations and terminations to determine
whether extended periods of time exist during which at least one aircraft is
always on the ground at a particular station.
It need not be the same aircraft,
only the same type. If a period of sufficient duration to allow a flight or sequence
of flights to be made that will return the aircraft by the end of the period, then it
is possible to turn the existing aircraft in such a way as to free one to make this
flight.
The final means of uncovering incomplete turns is to locate discontinuous
sequences.
When one flight in a sequence does not turn to the next one, this
discontinuity creates the need for an additional aircraft.
If the two flights do not
terminate and originate at the same station, it is necessary to either add a flight
that bridges this gap, delete a flight or reassign these flights to other sequences
where they can successfully be turned.
An additional validation is necessary to assure that all maintenance
requirements are met. Each sequence can be checked separately to determine
whether it spends sufficient time at appropriate maintenance bases. A predetermined set of guidelines can be established such as "each aircraft must spend 8
continuous hours at a maintenance base at least every other day." The schedule
-77can then be scanned for violations of this requirement.
It is possible that by
turning aircraft differently some idle time on one aircraft can be assigned to one
in need of additional maintenance time.
When all input data has been checked for accuracy and consistency, each
sequence has been completely defined, maintainence requirements are met, and
the proper number of aircraft are used, then the schedule is validated.
Although
validity checks may be performed periodically during the production and manipulation phases, they are not an inherent part of these tasks.
A number of
displays and procedures for confirming and aiding the creation of valid schedules
are needed.
4.2.4
Evaluation
The evaluation process may also be done periodically or even concurrently
with the other phases of schedule development. However, requirements of the
evaluation phase differ from those of other phases and they are not essential to
the tasks in these other phases.
Evaluation is the process of reviewing the consequences of the schedule to
determine whether it meets the goals the scheduler strived for.
These goals
relate both to the services offered and to the efficiency and economy of the
schedule. Service evaluation involves looking at the services offered in each
market and comparing them both with the desired pattern and against competitive
offerings.
Economic evaluation is done to determine if all of the airlines
resources (aircraft, station facilities, maintenance facilities, etc.) have been fully
but not over utilized.
-78-
Service evaluation is a particularly important function.
In order to
accomplish this function, services must be extracted from the network of flights
and crossovers.
It is difficult to evaluate the service pattern in a single market
from the complex web of services in all markets.
However, when isolated, service
patterns over a period of time can be observed and evaluated effectively.
When evaluating services, it is important to be able to distinguish clearly
between different types of service. A depiction of services is deficient if the
distinctions between nonstop, through and connecting service are not immediately
and graphically apparent.
It is essential that the scheduler be able to find periods
when service is either inadequate or excessive for a given market demand and
competitive environment.
Two items are particularly important to the goal of efficiency, these are:
station and aircraft utilization. In schedule preparation, the problem of station
utilization
can be reduced to gate utilization.
The concern here is that the
schedule require only available gates and that no excessive peaks of activity occur
at a station that would require large numbers of gates and station personnel for
brief periods and go unused at other times.
To make this evaluation, an an analysis of station use similar to the one
made for services.
All of the ground portions of each flight and interflight
periods need to be aggregated, one station at a time.
This subset of the full
schedule facilitates to observation of the activity at each station and identification of peak periods. This process may be done concurrently with the schedule
manipulation phase to deal with problems quickly.
-79-
Aircraft utilization is another important measure of the efficiency of a
schedule. Aircraft utilization is, of course, a discrete quantity -- an aircraft is
either flying or it isn't. Therefore, it is necessary to measure average utilization
for one or more aircraft over a period of time.
all aircraft
efficiency
A global average utilization for
during an entire schedule period is a useful measure of the overall
of the schedule.
However, to pinpoint deficiencies
more specific
averages are important.
Utilization can be calculated for each type of aircraft separately.
This can
be further broken down into utilization for each individual aircraft sequence over
the schedule period. Finally, utilization for an aircraft, a type or the whole fleet
can be computed for a smaller period such as a day or a week.
always possible that low utilization
on one aircraft
Although it is
or on one day is necessary
so
that aircraft can be positioned for important flights or maintanence, these
averages help identify weak areas in the schedule.
-80-
4.3
Requirements of CRT Displays
It may seem tempting to simply program a computer to draw identical
replicas of the scheduler's wall charts.
impossible due to the restrictions
However, this is impractical
if not
imposed by the CRT* display and is not
desirable since it would not permit the dynamic capability of the computer to be
utilized.
In addition to the many advantages of interactive graphics, some
drawbacks also exist. The size of a CRT cannot approach the size of a room full
of wall charts.
Screen resolution-the
number of discrete addressable locations--
cannot easily approach what can be reached with pencil and paper.
content and format of display elements must be deterministically
specified for all
possible situations so that the user not be required to complicate
supplying this information.
Location,
his job by
Each of these constraints must be considered in the
design of a suitable CRT display.
4.3.1
Size
CRT consoles range in size from a few inches to a few feet across.
Yet
even the largest of these can present only a function of the data shown on the
wall cahrts currently used. Of course, the screen can act as a window that could
be slid around so that various portions of the wall chart become visible.
The
problem then reduces to a question of how much can be displayed at one time.
*The term "CRT", which literally means Cathode Ray Tube, will be used here to
refer to any number of terminal screen displays that could be used with possibly
better results at lower cost. Other technologies include liquid crystal and plasma
displays.
-81The problem with this approach is that the limited screen space may
contain extraneous information while excluding pertinent information. There is
no guarantee that a wall chart can be drawn with all related flights on adjacent
locations.
To save space, displays for a CRT must compress the available data
into as compact a form as possible for the current requirements of the scheduler.
This is done by eliminating irrelevant
data, overlaying supplementary data and
physically compressing the representation
as much as possible.
Only those flights which relate to a particular
decision should be shown.
This subset of the total schedule must be extracted
according to some set of
unique characteristics exhibited by the flights and then only the necessary
information about each flight should be displayed.
Once the appropriate information has been selected it must be presented
by the most efficient
means possible.
This is accomplished
through three
mechanisms. Data items must be arranged on the screen to maximize the portion
of the screen allocated to actual data versus that allocated to spaces, hyphens or
other non-data elements used purely for readability.
Secondly, special symbols
other than standard alphanumeric and punctuation characters
can be used to
represent data in particularly compact forms. Finally, some information can be
shown in the same location as other data through the use of color, intensity and
other screen highlighting techniques.
A large portion of the wasted space on a wall chart is required to indicate
relationships between data items. Many of these relationships can be indicated in
more compact format that eliminate essentially uninformative gaps in the display.
One good example is the relationships between flights in the time dimension.
-82While it is important
to show events to scale along the time axis for some
applications, this wasteful representation can be compressed significantly. By
explicitly indicating the time of each event directly adjacent to those events, the
need to represent the time dimension through the physical location of data can be
eliminated. To further reduce the required space, the hour portion of the time of
each event can be removed and a single time scale placed at the edge of the
display. Even in highly compressed form, the explicit minutes and implied hours
are sufficient to uniquely identify the time of each event.
In contrast to the above example where the use of location to represent
data wastes space, the representation of turns can use location to save space. By
contrasting the station activity charts with sequence charts (described in Section
3.3) one can see that by locating flights that turn to one another in adjacent
locations, the wastful times needed to indicate turns are eliminated.
Special symbols provide another powerful method for reducing the space
required to represent certain types of information. Any data items that can take
on a small, finite number of values can be represented by a set of symbols with a
unique correspondence between symbols and data values.
For example, aircraft
types meet this criterion and thus can be displayed by a unique set of symbols.
The types of constraints placed on flight operating times--fixed, bracketed, and
limit--can also be indicated with special characters.
Although not available on all CRT displays, highlighting facilities such as
variable image intensity, reverse field (black on white), blinking and multiple
colors all serve to add another dimension to the screen.
used simply to improve readability.
This capability could be
For example, intervening spaces needed to
-83resolve the separation of data times can be eliminated by alternating the color of
the representations
to indicate separation.
A more sophisticated
lighting is to add information or indicate relationships.
use of high-
Members of a logical set
such as all stops at a given station or all services in a market can be identified
using highlights.
While screen size could be a very constraining
factor,
by carefully
designing displays using all available techniques, a CRT can become a viable
medium for schedule development.
4.3.2
Resolution
CRT display design is limited by another constraint
related to screen size.
that is somewhat
Because CRTs represent images with discrete points, or
"pixels" (picture elements), they have limited ability to accurately represent
excessively small detail. This problem calls for a number of design considerations
to meet these limitations.
The most obvious consideration is the size or scale of characters
represent data.
This determines the space required to produce a given image and
thereby determines the screen size required.
readability.
used to
Characters
must have sufficient
The limiting factor here is
must be sufficiently large to be recognizable and they
space between them to distinguish one character
from
another.
On many CRT consoles characters are predefined to be a certain size, thus
locking in their resolution to a given number of pixels.
Even without this
-84-
limitation, precise proportional shrinking of the screen image can only be carried
so far before it becomes unreadable or confused.
In fact, even if no single
character size has been specified, a minimum size should be determined and used
for all scales of the display. Readability should not be sacrificed for the sake of
compressing data into a limited space.
All reductions should be in the non-
essential space between characters.
Wall charts are often photo reduced to make them more manageable.
However, this is only done to realize the savings in space resulting from the
difference
in scale between the optimum size for manual drawing and the
minimum readable size.
A more important consequence of the resolution problem is its impact on
the complexity, rather than the mere quantity of data, that a display can have.
This consequence restricts
the types of display that can be implemented on a
CRT. Displays that involve complex networks of crisscrossing lines or many lines
that may run close to or on top of each other are difficult to produce.
In other
words, data must occupy unique locations on the screen.
If location on the screen is to be used as a means of relating information,
the meaning of a location must apply to only one data item at a time.
instance,
if one axis represents
For
time, then the events represented on one line
parallel to this axis must not be concurrent. In fact, events must be exclusive to
the point of not overlapping for an extended period of time--a time long enough
for data describing an event to fit in the space between prior and subsequent
events.
-85-
Format Generality
4.3.3
and procedures for producing displays must be
Since all the formats
preprogrammed,
displays must be designed that can be used in any possible
Some displays must be formatted as they are drawn.
On a manually
drawn wall chart it is not difficult for a scheduler to determine
how and where
situation.
new information
should be drawn as he prepares to draw it.
However, a
computerized version requires significant amounts of additional memory, machine
time and user time to permit the data to be displayed according to dynamically
supplied decision rules. Therefore, the design of CRT displays must be such that a
concise set of rules can be specified in advance to format and draw all of the
necessary displays.
To fully utilize the power of a computer, new displays must be drawn
frequently.
In addition, a number of different formats may be used to represent
the schedule information.
When drawn manually, the formats need only be
determined once for each data item. However, in the dynamic environment of
interactive
graphics the amount of information that would be needed to specify
all possible combinations and variations of displays could far exceed the data
required for the schedule itself.
If the decision rules for all possible displays were substantially different it
still might be possible to program the rules themselves rather than require the
user to enter formatting decisions. However, not only would the code be difficult
to completely specify and write, and even were it possible, the resulting program
would be excessively lengthy and slow.
-86-
Displays must be designed that can handle all conceivable possibilities in a
parallel and deterministic
manner.
Scheduling must not be complicated
by
artifically imposed constraints that force schedules to be designed around the
computer's deficiencies. Although this may require that slightly less informative
displays be available, the multitude of different views of the schedule that can be
produced instantly will more than make up for any losses in a single display.
-87-
4.4
Displays Design
CRT displays for scheduling must be designed to take into account all of
the complexities of aircraft schedules, the information requirements of a scheduler, and both the limitations and opportunities of the computer controlled CRT.
In this chapter a collection of many different displays is presented to satisfy these
requirements.
It is proposed that these all be used in combination, each display
providing a somewhat different view of the schedule that complements the others.
Three levels of display are considered (see Figure 4-1).
The first level
contains one basic display with no variations or options; this display closely
resembles the sequence chart described earlier.
From here we expand outward in
three directions, to the second level, by making three different types of modifi-
cations to the original format.
Modifications are made to the scale and
information content, to the order and arrangement of information, and to the
third dimension of CRT displays:
highlights.
Finally, in the third level, combi-
nations of the three types of modification are considered, producing some of the
more powerful displays of all those considered.
Level One: The Basic Sequence Chart Display
4.4.1
All of the displays presented
here are based on this one basic format.
Sequence charts were chosen because they provide by far the best available
compromise of all of the design criteria.
Sequence charts represent events along
the time axis in a very clear and graphic manner, which becomes extremely
important
-88-
I
/
I
I
LEVEL ONE:
LEVEL ONE:
BASIC DISPLAY
LEVEL
TWO:
MODIFICATIONS
REE
COMBINED MODIFICATIONS
FIGURE 4-1
Three Levels of Display Derivatives
-89when the scheduler begins sliding and crossing flights.
deterministic
format of sequence charts.
Equally important,
is the
No matter how complex or large a
schedule gets, each sequence of flight will always be drawn the same way and on
only one line.
Absence of confusing lines that connect sequences not only
eliminates the immense problems of reformatting
for each unique schedule, but
also saves space on the CRT permitting maximum compression of information.
Also, as we will see, the sequence chart provides a very convenient base
from which to branch out into various supplementary
displays.
Having this
common base makes it possible to reuse some program modules to draw more than
one display. It also helps the scheduler grasp the meaning of each display through
its relationship to other ones. Finally, it makes the commands required to access
these displays simpler to design, understand, and use.
The primary use of the sequence display will be during the schedule
manipulation phase, to visualize the relationships
between a given set of flights
in both the time and space dimensions. With this information a scheduler will be
able to make decisions concerning turns, crossovers and sliding.
Obviously, the
sequences selected to be shown must be related in some logical manner to one of
these decisions.
A basic sequence display will only consist of itineraries for a single date.
Once a scheduler's view has been narrowed to this point, further choices must be
made about the time and space relationships of interest for a particular decision.
For a turn decision information will be needed on those flights that turn at a
particular
station within some time period.
Since only flights using the same
aircraft type can turn to one another, this information should be specified to
-90-
eliminate irrelevant sequences.
A crossover decision will require information on
all flights using any aircraft type that have stops at a particular station within a
given time period. The scheduler may choose to further limit these to flights that
also serve a specific set of other stations for which connecting service is desired.
When sliding a flight, all flights that cross with it or turn to it or from it will be
needed so that these relationships can be observed during the sliding process.
Of course, a scheduler
may need to observe the potential
turns
and
crossovers at each station used by a particular flight, not only the existing ones.
To do this may require a display for each station for a reasonable time period
around the time the given flight uses it. By using these together,
a flight can be
slid to optimize the crossovers and turns available to it, and then these
relationships can be specified.
It can be seen from this discussion that the sequence chart must contain
the itinerary, operating times, time constraints and aircraft type. The important
relationships are not influenced by flight number, therefore, this can be excluded
for most purposes. It is also evident that the time scale should be such that under
typical circumstances
flights centered in the display can be shown with most or
all of the flights that turn to and from it. In rough terms this means that 12 to 24
hours is excessive, while one to three hours is insufficient.
time scale resolution.
Events occuring within a few minutes of each other should
be readily distinguished.
Finally, it is important
possible be displayed at once so that frequent
required.
Also important is the
that as many sequences as
paging of the display is not
-91-
The display designs discussed from here on will be based on an 80 columns
wide CRT with the standard ASCII character set.
Additional characters
discussed as appropriate but will only be presented as alternatives.
will be
Highlighting
with black and white will be assumed, including blinking, reverse field and two
intensity levels.
The use of color will be discussed as an optional feature for
improving displays.
To represent the required information, nine characters are needed for each
station:
three for a station identifier, two for the arrival time and two for the
departure times (where only the minutes past the hour are shown) one for the time
constraint for the ground time and one for the departure time constraint.
(Note
that constraining a departure automatically constrains the next arrival since block
times are predetermined).
In addition to this, the time axes must be labeled with
the hours. This could take up a single line, or, if more than one time zone needs
to be shown, as many as three lines may be used. Each sequence will be labeled
along the other axis with aircraft type when more than one type is present on the
display.
This label normally requires three ASCII characters,
but, if special
characters are devised to represent aircraft types, only a single character
location will be necessary.
Figure 4-2a shows two ways to arrange nine characters with no intervening
spaces. In order to distinquish between separate data items, and to indicate which
data items are associated, for example, which time is the arrival at which station,
highlighting must be used.
background:
In the first example, four colors are needed as
one for arrival time, one for departure, one for a station and its
corresponding ground time constraint, and the fourth for the departure/arrival
time constraint.
The second example separates arrival and departure times so
-9 2-
FIGURE 4-2
Alternative Data Arrangemerts
'"'
i····
.,,'''' .=,.
'''
· ..c
r'.·.·.·.
I···.
i·:· · ·
L·
r -1
1
=-X
.I -<771
A.
'""'
''' ''1
·,·I r
·-···· ·
'''''
I
I
Li
I
:,
.·
·.·.·;·::·.1:·:.
s ;· ·
''"''·2-·r,:_
·:..."
"·· · · · ·'·'·'·' "* · ·
:j
.-.;··.-.·.·r··.·.·.·
·
--,·.·.·.···. '·'·"".·.·.·.·.·. ;·:·:-:·:
.i
· · ·-·.(
C:·`.·.· -'.'.···
' ' ' ' ' ' ' '
""""'
··· "'
""''' ' ' ' ' ' ' ` ' ' '
:';·;:-:·:·:·:·:·f.I-·;';':
.
-,! l
c
-
Data Arrangement and Color Pattern for
Nine Character Blocks.
a.
------
,-----------I
A
LIX-
il-----f~
.-
t ------
i
,___
~____-.1.,,~~~.11~·ll--ILI-----_
1-·1·--^1
~-I...I-·
_ . .·- -----
.-
. -
-
~
·
·--
----
- -
-
..
I
,
1-
1
; t
li
I- -
·-·----- ···----- -·--·--
·---- --------f: r :,
·-·---,
11
11
?.
17..
.
.. ..
-
_:
;,
. 1^
.
_-1
I
-
.
I
_
b. Data Arrangement and Color Pattern
and Twelve Character
Blocks.
for Ten
-93-
that one color may be used to represent
the station, times and ground time
constraint, with a second color used for the flight times constraint.
Figure 4-2b shows an alternative format.
separated by the ground time constraint.
In this example the times are
This constraint is shown centered over
the station, clearly indicating its association with that station.
Flight time
constraints are shown centered below their corresponding times. In the first form
three colors are needed to distinguish the separation between stations since the
departure and arrival times run together.
This format uses ten character
locations, while the second, expanded, format uses twelve.
Some of these wasted
spaces can be used to indicate the break between separate flights.
bars shown in both examples accomplish this.
The vertical
-94-
The time scale can be varied signficantly,
but only in discrete steps. Each
character position must represent an integral number of minutes. Furthermore, it
is difficult to read a bastardized
scale where a unit distance (one character)
represents anything other than 0, 1, 5, 10, or 15 minutes.
Fifteen is acceptable
because it is an even quarter hour. Larger blocks such as half an hour or an hour
do not provide sufficient granularity to be meaningful,
nor do they provide suf-
ficient room to place the required data.
On a screen that is 80 characters wide, the four scales suggested yield the
following:
Minutes/
Block
Blocks/
Hours: Min/
Hours:Min/
Hour
80 Char. Screen
77 Char. Screen
1
60
1:20
1:17
5
12
6:40
6:25
10
6
13:20
12:50
15
4
20:00
19:15
The last column indicates the amount of time represented on a screen where three
character locations have been used along the axis to indicate aircraft type. From
this it is clear that the most appropriate scale uses five minutes per block since
this meets our earlier criteria for scale size.
Next we will consider how the station data will be distributed along the
time axis.
When drawn to scale, the data will not usually be shown at regular
-95-
intervals adjacent to one another as Figure 4-2 depicted them. Rather, the times
will be drawn in their proper location along the time axis to give a realistic
impression of the relationships between events.
Figure 4-3 shows a complete sequence chart as it would look on a screen
that is 80 characters wide and 24 lines long. This format is the central and most
important representation
of a schedule that can be implemented on a CRT screen.
It provides a good compromise in terms of scale, information
character
arrangement.
content,
and
This is a general purpose display; all others described
later serve more specialized functions.
The format seen in Figure 4-3 is based on the twelve character
shown in Figure 4-2 with a number of extensions.
display
First of all, two time axis are
provided along the top of the screen so that two time zones can be represented at
once.
Along the vertical axis are sequence numbers and aircraft types for each
sequence shown. The rest of the screen contains flight itineraries, arrival and
departure times, time constraints and a number of special characters that relate
information about the schedule.
Examples in Figure 4-3 show that, normally, arrivals and departures are
indicated by a bracket in the column corresponding to time of these events. A
two digit number representing the minutes portion of the time is placed above the
bracket.
Between a departure, or open-bracket, and an arrival, or close-bracket,
are a string of equal signs and the operating time constraint for that arrivaldeparture pair.
-96-
.....
C'
I i mc
,"I
l
0
C
C0
l
D
CO
k
~~o
~
~
II
II
II
--II
3
IIV / 1
II
II
II
II
II
Ii
II
II
II
Ii vII ii
II
II
II
II
~~~~~
<
Ln
i
--
t--I
C",J
II
II
Il
II
II
II
II
i-
II 0 --
II
II
-
II
II1
II
II U' -II
II
II
ii
II
iiI
II
(
II
II L
II
II
II
II
IIC 011
il C-
r-I r IIIII LLL
(0 Li-- *II
b--C)i
I
-0 I1
ry')
-
II
C'
--lrII0
0
0
i
l"" .... ~~~~~.~-~
I ·
iW
B7.
00O*
t
"
CO
cCc
I
~-.
~
II
Lr\O
L
0I >II -Iy =r -~u =f
(Y
II
In
I
0
O
C
II
pc r/>,V><
i
III
II
t
II
II
II
ii (,
II
I
II
II
II
II
X
>-4
avII
II
II
II
II
CII -
I
v~ v~
L
II "--
0
L.
i
II
ti`Cif III I. IIv
ICL
II
~
l"'-
i
XI
_
II ~O c
il
ii
II0O II 0 II L ~- II
II Lf-- L--
II
II
II
II
II
II
II
II
II
II
cI
i1
Ii O
iiII
<II
I
....
- M
I--
(
II--
It,
X
,
<
II X
V
~ X
X ,~
L',-·I~9 '- X
~ J>
X V
CI
a .-
<0I
0
Ln
a
a. 0
- L
-..
n
II
It<
II
Ci
.
II
U
>k
L-
0
CJ 1i
-
00
cc,
00
mo
C~
03
_-Z
L."
( J
Cc
I t
D~
itIt
II
-
(-c
r
oo
k.O
'
0
Ii
II
II
II
It it
It
I---
L
i~
I~
IiI~
t
m-
C
II
Ii ii
~/
I1
II
II
I
II
(" III
I1
a-
'..
II
"'I
LC N
0%
C
II
1I1 1-I
II
II
II
II
II
II
II
II
II
II
It
II
II J
II
IIIi
iiII L
IIII 0U ' III
Ii
LC-
iil
II
,
IiI0 II0
u
---II (l- r,
II ~J --
t\
It~
Ii
-<
LInLr)
II
II
I
> Lr >
$C
X
Ln II
--III
I NM
cc,
-
.C.
L~
u-....
"I,
II c ,
II
II
X
N t-l
0-uO
II N < J:
X II( I
C
00.... 0O0 * XX-- --II
L II
II
~
i
II
~ N
II
It
iiin
I:
II 0~
II
0
L
M
aV
~
III~
i c<I
~
N
: ¢ N
X V Ln
>) L-i: V X
LC
1- V (,.)
*
,
C
O
II
V
~~~oil
V ~oXO~
0000i
TCO
L~
II
-
-
(III 110.II''0 IIt0 !1I II II 0 .
·
O
'
· -O
CY)-
Ii
II
II
II
II
i
~
ii\
II
1_~' , II II
----,r·--OII
c
CVO
C ,
L
ItI
il
t,
iiII n
I
ry')~ 0 00
CXJ
IIOo
I
I
II
I
0
G
II
II
00 0-O0C00·O
-- i II IIr,II0 --c,--
C~
.
III
0 X
II ;
IILn--LC -L..f%--Lfl-OC (', O",,
M -f LC)\Do tc 0
I
It
C'J
t-*-0c30
j -, Z- Lf"'
- - O,\--
II
t
L ---
r(-'
r- c (J
-b o
e
a-
II
-
-
-
-
a-
-
C
(N r - Zr
(N (
(
'U
-97When a flight terminates or originates a verticai bar is used in place of the
bracket indicating the separation between flights.
If, however, a flight ter-
minates, but doesn't turn to another flight, a slash "I" is used to mark the arrival.
Originations that don't turn from earlier flights are indicated with a back-slash.
Ground times greater than 20 minutes and flight times greater than 15
minutes can always be represented with all of these symbols present and each
time aligned with its first digit in the block corresponding to its value. However,
when operating times are shorter than these limits, the arrival and departure
symbols can be dropped to allow events to occur closer together.
The stop at
station "D" in sequence 001 is only 11 minutes long. Stops as short as 6 minutes
can be shown by eliminating both the arrival and departure symbols.
Since the format was designed to make individual data items distinquishable without the use of highlight or color, these facilities are still available for
making the screen easier to read. Only background color will be used here so that
foreground color (the color of the characters themselves) can be reserved for the
highlighting function which will allow additional information to be overlayed on
the display.
Figure 4-4 shows how three background colors can be used to
separate the axes from the rest of the display and to separate each individual
sequence from the adjacent ones. This pattern of stripes makes it much easier to
read the display by associating related data items.
Table 4-1 lists all of the non alphanumeric characters
along with suggested aleternatives
used in the display,
that could be used if special characters
could
be defined. These alternative characters more clearly or compactly represent
their
meaning and can not be confused with other meanings of the ASCII
-98-
2 CHARACTERS
3 CHARACTERS
__
T
FIGURE 4-4
Use of Back round Color in Sequence Display
-99-
TlBLE 4-1
Character Definitions
ASC II
I
]
eaegin
o Segnen
ning
LBeginning of Segmlent
Alternative
t-
End of Segnient
A_
Flight
lMock
Ptinned Ti me Constraint
Constraint
Lower Limit Tin1me
r-
Upper Limit Time Constraint
A
Bracketed Time Constraint
Uinconstrained
lTivme
(a
Turn
I
Origination, Termination
-Um
-100-
characters used earlier.
Neither color nor special characters are essential to this
display; however, their use greatly improves its readibility.
4.4.2
Level Two: Modified Displays
4.4.2.1
Scale Modifications
As shown in the preceding section, three additional scales can be used with
the same format:
one larger, where one block represents one minute, the other
two smaller, where one block represents either 10 or 15 minutes. The larger scale
is of limited value in its basic form; however, in combination with rearrangement
and highlighting modifications it produces a valuable display.
The two smaller
scales permit larger portions of sequences to be shown on one screen and are,
therefore, quite useful. These smaller scales do increase the difficulty of showing
data in its proper location along the time axis; consequently, special techniques
must be developed to represent these scales. Beyond the mere compression of the
basic format, a series of additional formats that remove some of the data from
each flight can provide valuable saummarizations
of the complete sequence
display.
The larger scale permits the addition of flight numbers to the information
already contained in the display.
Although this is not particularly important to
the schedule manipulation phase, we will see how, when rearranged and highlighted, it provides important information for the evaluation phase.
In the larger scale the positions of the station name and ground time
constraint are reversed.
The flight number is added to the second row at each
-101-
station and both the arriving and departing flight numbers are shown at a station
where a turn occurs. This format is shown in Figure 4-5. Dashes are added to the
ground portion to help indicate its length in a way similar to the equal signs used
in the flight portion. As shown the minimum ground time that can be represented
is eight minutes between flights or six minutes for a stop, when flight numbers of
the maximum length of four characters
are used.
It is very unlikely that these
minima will ever be reached.
The smaller scales introduce very restrictive limits on the minimum
duration of flight and ground times. Figure 4-6 shows flights compressed as far as
possible without altering the format. As the scale becomes smaller, the rule that
all times must be shown with their first digit in the proper column must be
enforced, even for the most compressed case.
Figures 4-6a and 4-6b show that
minimum ground and air time for the 10-minute-per-block scale range from 21
minutes to 30 minutes, while Figures 4-6c and 4-6d show these times to be 31 to
45 minutes for the 15-minute-per-block scale.
Minimum times are dependent on the precise value of the times involved in
relation to the definition of the "no change" boundaries.
For example, if one
operating time occurs at the end of a block (9, 19, 29 minutes... or 14, 29, 44...)
then the next time, which can be as close as two blocks later, can be at the
beginning
of a block (0, 10, 20...
or 15, 30, 45...)
30, 40, 50...or
45,00,15,
respectively, thereby representing the minimum times of 21 or 31 minutes.
However, if the first time happened to be elsewhere in the block, then the
minimum time that could be shown could be as high as 30 or 45 minutes.
that two minimum times can not be found adjacent to one another.
Notice
-102-
co0o
N'
C3
II
II
11
II
0
L-·
co
1
II
C 1II
LrU
CM
~
'
.
-
OD
.
II
11
II
11
II
0
'")
CdJ
IINO
I
I 1 11
0
II
II
II
II
II
I
I
LfC
II
I
iI
CM
I
II
I
IIt
11-
11
.
.-
II
I I
it11
II
ItI
It
i1
iS
II
1l
l
IIl
it
,,~t
II1l
II
1
II
11
11
IILL
._
II
II
X
II
it I
<II
..
II
\
- It,i
II
11
I11
I
t'I
0CM 0 r-0
L
TIt
1
II,-i
I
_
-F
C)
It
II
III
Ox
..
II
11-
1I
II1
0
IXI
M11
1
0n
0'
o "-
1>CT_ I
11
CMC
P
0 O 0X
L3O
L
I
11
m0
CM
A
1 II II
-
C
,
,,
II
t1
.o
(
II
..
0 , 0 ,
CMO
1C
II
IIIIII
II
I
·JY)
..
IIN -
1
II
L
('x
OJ-
I
II
ko
CC)
II
It
1 ( II
( I
_U-.
"
itl C
I
0
I
·.
x)
-'"
=r
O
II
1
II
I
Ln
ts
II
..
O
-3
II
II
11
II
IIN
=
0O
II
III
I
"'
0",
II
l
I
----
M2)
L
II
II
0
03
C
00 '
II
II
II
1
0
t--' ~
0
Ln
O
C
II
II
II
II
~
0
I-~
3
CMi
,
II
0
O
O,\
II
II
II
II
I1
II
II
10O-
i
11
IC
II
11
I o'0
- N M T LC t-:CC)
00 C'O - NO
r-
-
'--
--
"---
'
C" CM CMNC
CJ
-103-
EST 10:00 11:00 12:00
a)
00*30 00*30
]AAA=*BBB=*]
EST 10:00 11:00 12:00
b)
EST
c)
09*30 09*30
]AAA=*]BBB=*]
10
11
12
13
00*45 30*15
]AAA=* ]BBB=* ]
EST
10
11
12
13
14*45 44*15
]AAA=*]BBB=*]
FIGURE 4-6
Compressed Display Scales
-104Frequently
30 or 45 minutes may not be sufficiently
schedule times, particularly for ground times.
small for some
Another technqiue is useful to
allow somewhat shorter times to be displayed without altering the basic format.
We can relax the requirement that each time be shown with its first digit in the
column corresponding to the time it represents.
What is absolutely essential is
that at least the first digit be in one of the columns corresponding to the proper
hour. As long as this condition is met, no ambiguities can exist -- all times can be
interpreted even though they may not be shown in their proper locations.
Figure 4-7a shows a typical flight represented
in the 15/1 scale.
The
minimum times that can be shown are much more difficult to specify in this case.
In the example shown, the flight begins at 10:00. The ground time could have
been as short as one minute because the departure time can be shown aligned with
the same hour as the departure.
A 20 minute ground time is shown, followed by a
one hour and ten minute flight.
This flight could have arrived as early as 11:00
because the arrival time is shown within the eleven o'clock columns.
Figure 4-7b shows a portion of a sequence in the 10/1 scale.
Because the
hour changes every six blocks while the flight times can be drawn every third
block, we see that the third time of a series always starts in the same position in
the hour as the first time. Therefore, the period of this scale is two operations,
one flight time and one ground time.
It is also the case that one period, being six
blocks, spans exactly one hour. Consequently, if any adjacent pair of flight and
ground times sum to one hour or more, they will completely justify the space
allocated to them. As long as every adjacent pair of operations together take one
hour or more this display scale can be produced.
-105-
11
10
12
13
14
a) 00*20 30*01 00*
AAA=*]BBB[ ^ICCC
10:00 11:00 12:00 13:00
b)
00*20 00*15 00
]AAA=*]IBBB=*]CCC
10:00 11:00 12:00 13:00
c)
58*59 00*01 00*
IAAA=*]BBB=*ICCC
10
11
12
13
14
00*30 30*00 00*
d)
IAAA=*]BBB=*]CCC
FIGURE 4-7
Altered Format Compressed Display Scales
-106A single pair of times need not always be one hour long. For example,
Figure 4-7c shows an extreme example of one minute flight and ground times.
This is possible because the times happened to fall within two hours and so the
first two could be at the end of one hour while the third and fourth are at the
beginning of the next hour. Notice, though, that because these times are so close
to the hour, the preceding and following pair of operating times cannot be
compressed to this extent.
It is the value of the times involved and their location
in the hour that determines how much they can be compressed past the nominal
limit of one hour. One hour is adequate for the vast majority of operations, since
ground times under 20 minutes and flight times under 40 minutes are rarely
encountered.
Figure 4-7d shows a portion of a sequence in the 15/1 scale.
In this case
the period at which hours and flight times repeat is three hours. Each three hour
period contains four operating times, two ground times, and two flight times.
Again, depending on the values of the times involved, one period could span as
little as two hours and one minute. Although it is more difficult to specify what
combinations of times can always be shown, it can be guaranteed that schedules
containing no ground times less than 30 minutes and no flight times less than 60
minutes can be displayed.
Many shorter
times can probably be displayed,
depending on when the operations occur.
Even though most of a display in one of these compressed scales can be
drawn, a few stops may not fit within the limitations
described. If this occurs,
rather than not using the display at all, it can be abbreviated further to
accomodate these special circumstances. If one time cannot be shown under the
proper hour, this time will be removed from the display entirely.
This will allow
-107-
the adjacent station identifier and its times to be moved closer to the abbreviate
station.
To indicate this loss of information explicitly, a single question mark or
special symbol, will be placed above the station identifier where the time would
have been.
To simplify all of these scales it may be desirable to eliminate the full
itinerary and its times, and replace all stops but the origination and termination
with just the flight nimber, as shown in Figure 4-8. This representation
is still
useful for turning flights, since the intermediate stops have no effect on turn
decisions. If it is necessary in the smaller scales to compress the stations further
than the flight number (which could be up to four characters
wide) allows, then
the flight number will be skipped.
Any of these scales could be used with a combination of flights represented
in both formats.
For instance, one flight might be shown with its full itinerary so
that it can be slid, while other flights that it might turn to are shown with
origination, flight number and termination only.
No smaller scale could accurately represent time information.
number of increasingly compacted
However, a
displays are very useful for getting a broad
overview of the current status of the schedule.
From these a scheduler can find
those portions of the schedule that need more work and then blow them up to a
larger scale.
To create these displays some information
is removed or sum-
marized in more aggregate forms so it can fit in the limited space.
The first compressed display not showing time is depicted in Figure 4-9.
Sequences are shown arranged by aircraft type with those flights not assigned to
-108-
1 ,II
F ~~~~~
II
CN~--,,
0
,,
s~c
1.~
C
JC
)0 C
Cl
tr
,,
.,5
II(%I--
O O' -- C
..
....
M
C*
~- Cl V L
-- O
- -;
__
--
\
61
C
O
0
M
p0
n
CO
o0
00o0 0o
k12)
ohn
J' /X
CI
-- J3I. --II L~
O0 [-- <(- \/ X
X
Lr
3: '
rOJ
GO
Co-
E-
....
3
00
O 0
S
C
0
CD
.n
\t
....
Cso
C<
_:r :
03 U)
11
11
1
II
11
0¢
E
VD
CY
I
C
11
D
11 r- C3
L I :-- I
t~--
St=t
__
tr)
0o
OL~
0n
0~D0
r--,r--'
.C
co
Om
C0 CD
uU
)
t1
r x0
~..*S) .
11 o
LDZ
l
--11
0
0n
MD
'
Cdj CT~~~~
Oh
-Ln
\0
M
cr
C: 0
OOLn
M
J
L-
1
Lr
~~O
0 0
OdJ
-
,
%
r
C
O
1
r
c:
C
--II1 -II
II
c~~-II II
0
-- O-O0 011
CI0
r-C
r
-- 0 I1
O o
O *
[... ,-O~
O
X (C
X
MX
M'--'O
04 --
C
c
c-
--
$
4 -(
1
:<(c
0--<II Oh J:>...
kl~ c;;
('~ --
O
Ln
L
O O Y)C c -~
00
3
Cm *sC
~
a:
<
O4O,
Lr
11 ._
CQL
II
CEI
II
O
II
II
O
0C'kJoO 0--O
.t -II
II
II
C~~~
II Ln II
L
C
m)
C/D
<
CR
oO
\/C~
11
.......
II
N
0
*...O
cO
P
c
CIO
1
I
0) '"" --OO V
0C
Ur_1
C.,
r ' ~....
(J
II
J f" L--
00,
0'0
C
CO
1
a
11 NII
L/
IIO
MJ)
. O[ O
\m
cA0
II
cD c---
O
CD
'
II
II
....
O
iI
11
0 0
co 0
II
1If
II --_
OhIC/
I1
N 0o 0n NC c
Ev(
CD ~rEO U O Uc- O4
JU O a o GO -
00
oO
'\J M J LD DhO ,-.-C.r' J-...L "O - DoC
O
- -- Cr
r- - - --
(. rm J.
CJ
N,
-109-
CO
LTI
o
co
o
E-
-o
O
b-
0--
XL
X
~
0 Cr> X
-
"
L
X <O
--X
O
M_
MI
C
rml
co
o
a s
O
,0
m
-i
E-
-C
:
C
-Cy
_
O
C"
L) O
NL
C0 5I 0o X¢:4
,x C0MC
C
o
L
c :
:
U C
E,-
:;
X> -
oo
cGr>- >
¢: X CL Cf >¢
Q)_
:r
cU
X o m~~~¢ C3
Cs [°
D
Ch
CM
0T
C0
fe
C
t
~-
-I O . .. - .
CJED-
OO
L ,.
r O
0000ro000000
..- C M
_
-. X C"
-00
0M.><C~
O O CO[-0-CO
' C
3= CT
,_
E
C
OCl:a.O XQ u>MC
,
(X
CM
.4
cOJ
X CMCM
m0
O--
O,
'0 C
I:
>- L)N.rCM
n
u
O
Z
0
TUN X
(\
-O
C l.
-
>
q X
UM-:>r
-
E
E
L.x E-.
n
CN1 - C
C
o.
O
00
nl
CT
X
O
or'
cC
._
.
y m
CY -- X m
\
Ln O
c
Om
X
X~.C--
s
Cz l r,o r..)
- 00
-CM
aa
N
M-L\. L-cCOC O,0- OC -MC rn
L D b- 00 oC
- C CNC C C
-- - -
-110sequences shown as a pool of flights following the sequences for each aircraft
type.
Each sequence is labeled with a sequence number, aircraft
type, daily
utilization, and average stage length; then, a series of alternating stations (which
represent the originations and terminations of flights) and flight numbers are
shown.
Normally, a station that
is the termination
of one flight and the
origination of the next flight, to which the first one turns, is only shown once.
However, when two adjacent flights do not turn to one another, the origination
and termination stations are both shown and a slash is used to separate them.
In this display only a single day's flights are shown. A flight belongs to a
given day if the departure time in the time zone from which it departs is between
midnight (0 hours) and midnight (24:00) on the given day.
based on this same set of flights,
Daily utilization is
even though the elapsed time from the first
departure to the last arrival may be well over 24 hours. This definition is useful
because it associates a unique set of flights with each day without specifying a
reference time zone.
Although sequences are normally shown with stations and flights in the
proper chronological
order, two exceptions are made.
First, sequences that
require more than a full line on the screen are simply continued on the next line,
rather than on a second page.
Second, those flights which do not belong to any
sequence are grouped together into a pool of flights that are displayed in order of
increasing flight number, with as many to a line as will fit.
To further compress the display, each sequence is shown with the same
label, but now only the flight numbers are shown, without any station identifiers.
This allows the screen to be split vertically in half, thereby doubling the number
-111of sequences and lines of pool flights that can be displayed.
Figure 4-10 is an
example of a portion of such a display. Slashes are still used here to represent
discontinuities in a sequence.
This display can be compressed one more time before its content
dramatically altered.
is
Figure 4-11 shows the screen split into three columns.
Utilization and stage length have been removed and the aircraft
shown once to label a whole set of sequences.
type is only
Slashes are used to show
discontinuities and sequences exceeding one line are again continued on the next
line. In both this display and the previous one described, a single aircraft
type
will be continued in the next column before starting a new screen.
While these last three displays provide a valuable overview and index to
one day's schedule, summaries extending over a week or the whole schedule period
are also essential so that the scheduler can identify those days that require
further work. Two such displays will be described here. Both of them are merely
inventories of flights including only such summary statistics as number of flights
or discontinuities per day, daily utilization,
and aircraft requirements.
Figure 4-12 shows a sequence inventory.
Each line contains one week of a
sequence labeled with its sequence number, aircraft type, and weekly average
utilization.
Under the column for each day of the week is the daily utilization and
a count of the number of terminations occuring in that sequence on that day that
are not explicitly turned to subsequent flights. More than one week of a schedule
period may be shown by continuing the sequence on subsequent lines.
-112-
co o
3CLn
t--
m
00
C0
0
rM
-
Cj
L
O Ec Ch-- =rC
O
rvb~4
LO
Ci
_c
,
C\J
-
O
C
O
~On
1O
tC .
N
C
E
-DM
.
*-
o\~~m
0
co
L
C3
C
C
c'H-*-00CI
InO[
0 o0 m o
CT :Z~- 0 0
co
io
&:-
o-
0
0
CM
0C
ov
-
C
,
.LnO
E
c
0
-
cc
,0m
.CD
CM U
C
'
2
i
-
E
C0
000 00,-C-O
--
O
00
CL}
co
t
'
W
0
,''--
03,'H
~e
L C.
e
t--
) C' a,
UU
0' 0.O'
UU
U')
<
m
Cr0- CM rrf
CM
e
-
::D-
co
\0
~
oc"
C0
O~N
Q
1= C)
L-100
C'/OO
'-CM fv'
9= l
00
00
- D t--COoa-O
C - COM
0
'
q-
'-,-
Ln
UO
bt-cc Cr.O -) CJM
rN =---- - --C C C CMC
N
-113-
00
C)
JU
Lr)
M
c~
f0
E
02
a)
,0
z
Ln
z
00
Lr)
0 O 0
M
CM
C-
L-'CM
O '-O
00
c
'.0
LC1
CM
C\j
C'
LI
U",
'.0
CM
00
CM
1C0j
cl,
\0
Lr)
M
CCl
0
0
e4
--C\J
ry)
C) C %
Ir0'-00O
u*0-C0
q%-
1-
1pc
q
\
E
1
I
c
J-) C
I
Ot t,~
I -~ L
r- Cl 00
II 0 O-OOI
I 0--'0
I ---
-'-C
OJ
CM)0M
C
_
O
O OOC
ON
C
I r CM
CM
-
C .o-
V-
-O
C
-J0
M1
- -
r.
C
'-0
C
N.
II C O'
'"- tCkCv
O J,C ON
t." . --M vO C'-'-0C
CM0000
00J-'
II -
L)
00
in
L
V
0
0
Lr
L1
O L)
~
00
0_
C_C N,
.N.N
oo
M
Lr)
0N
CMrY
- U) C Mr
O
00
\.O t
CL
L
LC
0 000
II _MC--M
Cf--CCUCM
MJ
JMM
L 0000w
0'~--CU0CU CU0OC
CU
~O
CU
MNC~
~-a
cmv-0 ) y
I----- …
…1'-…-rC,-…-I
N.
0- CM
L \0 t-c00o
I
.CL.l
C~000000000 000
CM
LI'.0tpco0 0 - C M
1-1
W 1-
LUn
L
LU1
-
--
--
\ssN.
O 0 -
1-
t-
1-
T-
- co C
ru
1-
1-
1-
I-
I-
--
C
-
CM
11-
CM
CM
1-
--
_m
1-
1-
%-
C M
r
C
-
01-4
a)
0l\
- C)
_
00-
--
'- t-
,-- 0
I
C
CM
U)
Md
9:1
-114-
0 O
c2~
o
-
IOo
¢kl
0-Q'-O
..0..
u0"-
bO
Cj
O
-"0 Cr" Oc0 0
CN
J
,---- -- ---
00
C
Lr\ O0o0
- %0,C\Ju0
fr~
Nf
o,,
. ..... kl
**
..........
C'.ioh
CT)
\ oC i.D
Cr) 0o
C) o~
Cl O
o~,o
.
c, C'0
D C"0
o
\0Z
ON0J*-
'.L0
.
H0: o 0o
O
0
C0L..C7U OD
*r
M
,.O 'O O O
""
x
-- -.n
'T--- L:-
C\j
LO
>
xcn 0'O
\Z
00
0
0
0
C'.
I
*
0
C c fm
o
O
"7
..
*
CC
)
mm
oC",
O
oo
C(J
O
O O
C O OO
M O O ---
-
\
J
n k0jD3r 00o oo0--b- 00 0 0J
C0)
0
LC)
J
--
C)
.~
CROOO
OD
on
0C)
_
HE;
. O M
r~t
~O
,
Ln
M
OD
_
c0 m 0
LN
~- OmO)
rr)
xro
.
(N
C
.- LnO
M
.(
oo
co_
0HF
.
0 L 0 0o oo
&m
O
r- C
o 00
C-'
C'
C
O
0
00
m -- U) D t OD C"\ Ctv-
C M
-
T-
T-
-
V \
t- OD
T-I-
-
-
, CD-
C
C
N M -C
C\
Cj
-115-
For each aircraft type one line is provided to total the data in each column
and one line is provided for the pool of unsequenced flights. The total line
contains the average utilization for all flights using the given aircraft type that
have been sequenced. Next, it has the average daily utilization
and the total
number of sequence discontinuities for each day of the week. The line for pool
flights is constructed somewhat differently, containing the number of individual
flights in the pool for each day and the count of the minimum number of aircraft
that would be needed on each day to fly all of the flights, both in sequences and in
the pool, if they were sequenced in the most efficient configuration (FIFO at each
station).
The final display is just a summary of data for each aircraft type for the
entire schedule period.
One line contains:
minimum aircraft requirement,
an aircraft
type identifier,
the
the number of sequences defined, the number of
flights contained in sequences, the number of continuous pieces of sequences, the
number of pool flights, the average utilization of the defined sequences, and the
average stage length of all flights. Figure 4-13 is an example of this display.
This whole set of displays can be used to evaluate the schedule, identify
weak or invalid portions, and then select the sequences and flights involved in a
given problem.
These flights can then be shown in one of the larger scale displays
and manipulated appropriately.
4.4.2.2
Rearrangement
By simply taking the data as it is presented
in the basic format and
rearranging it into nonsequential patterns, more information appropriate to a
single decision can be gathered into one location.
These rearrangements
take
-116-
oo
cr
C-'
co
CM
Ln
CMj
0',
DO
o0
CM
'JrCU
L.
0
CM
¢ CC
C'
0
C
MO
.-- .. ..
c~00
LN
.Z O
:O cc
0
ZO
u
Ex C)
L
CJ
HC'
0,
O0
HW~
, C
rM--Z U-~.O t.-a
C3
a0 0
I
r-
M
- T- - T - 'r
U')
LZ -'D ·-- OD aN Q
- -I
.
-
CM
CV
c- CMCMCMCM
-117individual stops, segments and flights, sometimes even from different days, and
collects them. Each display will center around a single station, market, or flight
and will show all related stops, segments, turns, or crossovers.
Turns can be viewed in two different ways. We can look at all flights that
turn to a given flight on all of this flight's days of application; or, we can choose
some period of time at a station and look at all flights that originate or terminate
at that time and place. It is clear that the first case requires a rearrangement of
parts of sequences from a number of days.
The second view may also be
rearranged to more conveniently represent the turns available.
An example of the first turn-display is shown in Figure 4-14a. The flight
shown at the top is the one chosen as the focus of this display.
The time axis is
broken in the middle of this flight to compress it into a shorter space. Below this
are pairs of flights with which the given flight turns on various dates.
Only the
flight numbers of these flights are shown, rather than the full itinerary.
When
both turns of the given flight are with the same two flights on a number of days,
the flights are shown only once.
Such multiple turns are labeled with the
beginning and ending dates and the days of the week for which that turn has been
specified.
Each different pair of turns is shown in chronological order of the first
day of application on which those turns are made.
Finally, a list of days of the
week for each week of the period on which the given flight is applicable, but has
not been turned, is given. This points out the need for additional turns on these
dates.
Figure 4-14b is a simple sequence chart with a set of sequences that have
either an origination or termination between 8:00 and 10:00 at station A. In this
-118coo
o0
oO
O
xD
J
C)
co
-- O--0--
0
-0
·
Ckl
C'
II
II
II
II
II
II
II
II
II
0
0
-l O
O
o
0
11
1It
f'r
O0
C
C-
CLrL-- -'-
000
.011
II < O
C
(
¢u
o
LFoo{YwC'
<
r
U
,O'-C
Dd
N
cO
CY'
Ct
CU 1 H
1
e
0-<
oC11
<
cJ
H
~L,
U
aj
II
H
0)
0
It11
I
II
It
11
It
II
it
II
~
II
II
II
II
II
c
Z
II
II
It
cuI
0
(
II
II
II
0n
O
--
II
\0
o0
X
X
Ln
m
CU
X C IIC
O
Cl)
I
Z
:
11
O
><
11
,J
L.
It
O O
O
---
D
0
LT
-- L-- O
<
"
1
II
II
O
IIi
MO
'0
CJ
011011011
L-'O0
o
03
-C
O'
J¢\
r)
I
(\
II
1
II
II
II
II
II
C
-
_II
II
II
C\j
ID
O
II
X
X
S
'M
CH
Cu" 0Co
OUOU
OUC
O C O E: O
-- C M'-
.1):Db.o0
0v- C
OO'"
-
-
LU"Dt- COONO - CN --
.
-
-
-
-
--
-
O
CU
Cu J
Cu Cul
-119co0
0
co
CL
o
o0
cQ
..
.,
C: a
C) J~
t'-,-
C~
kO
C
0·
III
X
1
11
X<III11
1.*- X
* I----- - 0YI -
0O
-
O
00\~
~
LF-
III
II
II
II
II
II
k,00
k.
C
C
O
a
m
*·
II
I
II
C
C\J
CM
OJ
Ln
CO,
II
II
IIC
II
ryc
aj
- Ii -11 I I
II
C-
b-II
LU
II
(' o
"
CO
-I
11
I0
It
C
..
(,
,
II
II
II
o I
riII
X
'11
C
11
1
1"
II
I
)
o
II
II
II I
{,/'
II
L'"-
'-
II
.
II
{'v'h
><
XL>< II
I
II
j
Lrin 11
CU
CC
C"O
O
'
*0
CU
./
It
ko
J
E
C
X
I
C'\
II LnII
tI iu-II
¢
C(---- ¢
it--
t
II
II
II
II
¢
¢
rr<
II
C
I
11
.
Lr'
'
II
II
II
II
II
c -)ii
I---
,
II
II
II
II
II
II
"
co
II
IItin-II
II
II
II
O
o
II
V-
'-s
C
w . UL)OUOUOU
o ao a o a
"- C
'Y
Lk.O
' -0
C)
a
C'CC-
CUM '- ir.O
--
--
-
T--
-
- CO ',C
-
-
T--
CJU
C r
CU C
,
CJ
-120-
CO
o
L" -
....
klZ
cm
('% ''-
C.
t
M
C
v
~U
II
II
II
II
t-O
CO
o
kD
.~
..-
-
O
II
II
X
II
II
II
II
::
11mC
X
II
II
o
-X
II
II
II1I- -II
oo
.-
II
'U
CM,--
,l--
II
II
II
I
II
CM
II
II
II
II
I-
I1
II
II
II
II
Is
II
II
II
'.0
C0
-
r
II
I
II
II
II
O
00
O C1
00
M)
O
CM
-
5
II
o 0\'D
0
IIL. Ii
,
L
II
11
II CY_
oII
M
'
%)
-
0
-
00iC
I
0
'.
-C
O
is
II
0
4-,
C
L
_
O--
II
C
'-
C so
'"- -
C
OC0
0
r
It
I1
I
II
-
'
11
II
11
n0
II
L-
I
CJ
II
11
II
y)I
0
c)
g-
0
(-)
II
II
X
II
iX
II
II
II
X~
Ln I
00
C
an
X
_
X
v
X
II
11
II Ln---
CM
0
._
il
rm
V
i
0
L
b"- \D
11
ilX
11
kC)
"
LC
'-
D
I\i
M0
- O-I II-II
M
;
11
Lt'C ) CJ
IJ '--CO
)
--
cO
C
O
II
00 011
O -0\
0C
IIL
II
II L --
II
1 II
CLn
II
II
II
0
II
)
\
CM
0
II
mi
m
\
\D
3
~
L~
i-
C-M
C
Cy)MvLn '. t-IC C O
-
i--
-
m0\
-
-
o
t---
--
'\o
0
C r
C C C C
M
-121display it is possible to see all turns occuring within this period as well as flights
that could turn during this time. Figure 4-14c, however, shows these same flights
rearranged so that the minimum number of aircraft are used.
The earliest
termination is shown followed by the earliest origination that departs late enough
to meet the specified interflight ground time constraint. All flights are shown in
this manner unless no flights are available to turn either to or from it, in which
case the flight is shown on a line by itself.
Flight numbers are added to this display to allow the turns that have
actually been defined to be traced, if necessary. All flights terminating at the
station are labeled with their flight numbers along the left side and, similarly,
originating flights are labeled on the right.
At the turn station, the station
identifier is not shown; instead, the flight number for the flight that the
terminating flight turns to is placed on top of the flight that the originating flight
turns from.
In a slightly modified form, these two displays are also useful for making
crossover decisions (see Section 2.1.3).
Instead of displaying only those flights
that turn at a particular station during a period of time, all flights that stop or
turn are shown. As seen in Figures 4-15a and 4-15b, the full itineraries are shown
in both of these displays. The large number of flights that will be included in this
display can be limited by specifying that only those flights that also serve some
set of stations, for which the scheduler wants to provide connecting service, be
shown (see below, Section 4.5.1).
Similarly, a multiple date display can be
produced showing all flights with which a given flight crosses on all of its days of
application.
-122C
o
00
o0
~io I~
~ ~
'.0
C.)
.L'%~
Ii
II
II
II
II
II
II
II
II
II
It
II
II
II
II
II
II*
II
II
II '. L
II C ---
II
II
((
CM
X
X
II(
(
L
II '. C
L"~-C~
I
00
I.yr,
II
II
II
II
II
II
II
00
O~
....
'.0
II
II
C
II
Un *
I
.
-
cv'h
II
It
L
0
,o0
C0
f~~~-C
II
L
J
~
~
m
c(I
II
~~~
It
It
CI
II
( [..~
oo
II :
II
II
II
II
II
II
II
II
II
I1
(
II
II "-
II
II
II (r)-II
II
II
II
II
II
Il0c
,-,
II
It
II
<
C
II
II
It
II
0
CD
LC0
kOT
II
II
~ \0
r -
II
=r-
II-.~ ' '
II
i
II
II
.. .. II
C ~
rw' 0
'.I
00
O0
CO
0'~
II
II<0 '
I
0
- \-IL
C
II
II
II
00
X
X
0
CY~
X
Cv'")--
CM
It l
II
C-
-
0~
T-
It
II
r-n
I
I
I
I
II
II
II
II
II
It
I
II
Lo)
+-
,--
U >(':
1
0>
U
I
I
I
I
I~
._~
I :
II ' II
I ~ I II
~iI
I II
III
= 0~~~~~~~~~~~~~
II
O IIC (
---
<
I
II
II
II
II
X
I
II
$C<
1:~
:
a
II
II
II
II
II
1
II
L--
II
<
II
II
II IX
("--
1:LrL
II
II
II
II
II
II
II
II
00
C'N
0
'3
CtIIaI
I
II
II
II
II
II
< 0
OIl
*
0'
C
-,
..
II
II
II
II
-,
-
t
1
I
0
I
I
Jl:~
J
I
L-ZIM0 II 0II
V
"-q
L
II
II
O~rrl
-- 0
'. 0
II
II
a III It
C 1(II 0v~Ia.
III
00011
....
---0
CO
IIII
~-
..~- .
CY"I
C*'J
II
CII
(
U
I
u IC C) I
II II
Ii-CMO
I II
II
II
C'r- II
II0
III
II
I 0
r~0l
I
III
L J(
i
II
II
I" ,C II
It
C
IIC - j II < I
II
II
CIII
II
0I I
IIO I
0~~~~~~~~~I
II
<.
IIII
II
II
t--k.O
L
II
I
L
C\1~~~~~~~~~~~~~~~~~~
II T0'\ 0 0o 0 II
N,
('X,
II
-
110 II
0
\D
II ~I
Ln ("Y'I~
C) C) ~--0k--tJ,--cvf~
-'0
II
'.)O0
It Lt
<0
m Cr -'
c~
0I
I <II
< <
III
co
~
II
1~
M
II
X
II C\J
C rI.
II
<II
a
II
m it -C
I
0'
·
O
II
E
I1
iiII
VEII
C
0L 0r II
N
Q
L.-
II
II
C
It
C
II
0IIa:u-J-LVVCa;
00
'IO
Il
II
II
II
II
(
It
i.CO
f
II
II
it
II
Ii
'.0
II
Z
.Il ( C
II L~ a
II 0 i
II
II
II
II
·....0
II
(
-
ji>-r'),.J
C L-J
00
m
0 < I:Zm
0 0 0 ~
L
W
II
II
II
("' .. J
II 0
II
II
II
II
Z
ItN Z
C
II
II
II
CI
'.0
II
CNJ
(
II L II
IIY'l.J
r(_) U
L)
II
II
CC
C)0"
vHT-UH*-L) X
L U 0
C-Jr'-
'--r
0
- U0
0 ~v-~
X
L' \.D t-C 0 C O v--
t/)
CU
-CM CM
N ry'-
L D -O
Cb-C~ --rr--v-- -
'-
-
0\0
C
-C
v-
- CMr' --
CMC C CJ
-123-
con
oo
co
II
(
O0
:,
.n~r -IIO~II
I1L
CM
II
II
L'
Crh
0m.
o
00
... ·.
11
II
0
11*
1
II
U
~
II
~no
L
E
II
II
I10C)~
II
II
II
II
"
II
I1
,II
V
",,
f)
CM
1
II
( II ~
II
O
l
" CoD
1... .: <I
0
O
II
II V
II
II
I_
I1
vIn
t
,
1
11
IIr
11
II
I1 II C@
II
II
II1
_
- ¢III0 II
C"
C
Cx
-
'
00 00
CO
O
M
,
0m
*0
LN
'-- 11
. II ¢
~
00
00
....
cO
wo
,00
CM
L5), v
ckJ
,r- '
II
II
( IIV
111
11
11
11
_
C
I
II
ii 0 II
11 0--
X
>
1<(
X7Ln
110
OII
1
C,)
li
11
,
II
i
II
C
er
II
11
0
O
O
t-- J--
[.- N
U
-
-O Ct
U
oC
aC o0 -t-- ,r
o.L-)r,C
0- t-- -'" --- d a)
iC:::
r
-v-.dL
>
II
II
In
N
--
3
0
CO
I
11 v
il L
11-
LII
*
L)
II
III
II
II
..11 1100
II
¢I III fo) | N1
sII
Lr)
CR
Cd 0
C
II
IIi
c
C
II
.0C. II0
II
U
- v" v , II
IIII
('x
-O0)
>
,
1
cc)
oO
I
II
II
II
II
II
ODt-0C
CO,,--
C (I-r
,--,---
C' r
L \D t- co c'0
--
-
-
CNJ C
C\J C
Cx
-124-
These four displays are very valuable to the scheduler.
They allow the
identification of turn and crossover opportunities and observation of the potential
impact of sliding a flight on these relationships at any stations or on other flights
that are involved.
As was demonstrated earlier, it is important to see the activity at each
station during the evaluation phase To do this it is useful to extract just the stops
at a station and display them in a nonsequential format.
This produces a Ramp
Chart; however, a number of problems are encountered with this display. As seen
in Figure 4-16 the flight number is used to label each stop and both the flight
numbers of the terminating and originating flights are used to label an interflight
turn.
Therefore, stops may have to be as long as 25 minutes and turns as long as
50 minutes to be displayed with the flight number. The second problem involves
the spacing between adjacent stops on a line.
In this display each line can be
interpreted as a physical gate at the station, and the separation between stops as
the time required to roll one aircraft
out and the next one in.
display is drawn some minimum "intraflight"
So when this
This
time must be specified.
minimum time must be greater than 10 minutes if all stops are to be shown with
at least one space between them.
Both of these time constraints
may be
bothersome which will lead, later, to the suggestion of a larger scale for this
display.
The final problem with this display involves the representation
of discon-
tinuous sequences. If a flight arrives at a station and never turns to another flight
it will occupy a gate,
and therfore, a line on the screen.
To avoid this problem,
all unturned flights will be coupled on a FIFO basis, with at least the minimum
specified interflight time between them. These coupled flights will be shown with
both
flight
numbers,
separated
by
a
slash
to
indicate
the
fact
co0
ol~
..
o
-125-
~co
Ecc
~C"O,
0
Ln
.Ln
0
O0
c3
,o3~
O
Ln
U
0
310
O""Lr
0
i~
O
0u-{
0n
LCO
,"C,O
cM
0
-
rCD
J.
O
n
0--
r0-
O
m
O -O
0
O
"
L0
0-
u
00
c
fJ_
Or-L
c
O
00
CD
O
0
Ou
- CM
- u-i
U " ca
(-CC
0r'-~ C, M'
'
tL- C 0 v- - v-- - - - C
m -J
C C
-126that they do not turn to one another.
Of course, the minimum interflight time
that can be displayed if the flight numbers are both four characters is again 50
minutes.
4.4.2.3
Highlight
In most of the displays described so far, the use of background color or
reverse video would aid in distinguishing between adjacent data fields of similar
content and improve readability.
Highlighting techniques, such as changing the
color or intensity, or flashing a character has a value beyond a mere improvement
in readability.
Highlighting can be used to locate similar data items scattered
around the screen or to single out one, or a group of data items for some kind of
special consideration.
When using a multiple color CRT it is best to divide the available colors in
to two groups, one for backgrounds, the other for foregrounds.
Background colors
should be selected so that each is easily distinguishable from the others and should
be a pale or "low temperature"
temperature"
color.
Foreground colors should be bold, "high
colors that stand out and can easily be distinguished from one
another against any of the backgrounds.* The meanings assigned to foreground
and background colors are completely independent of each other which avoids any
conflicts or compromises due to the colors involved.
When assigning a meaning to colors, care must be taken to assure that the
data items that will assume particular colors are mutually exclusive sets.
For
example, assigning one color to flight 101 and another to all flights that connect
to 102 is not allowable because flight 101 could conceivably be a member of both
*On the ISC Intecolor CRT, green, yellow, white and cyan make good backgrounds
for foregrounds of black, purple, red and blue.
-127sets.
Two colors are easily used in a binary mode, where one color represents all
items that meet some condition and the other color represents those that don't.
For more than two colors to be used, all data items must be separated into more
than two mutually exclusive groups, which is not often possible, although some
examples will be discussed.
If more than one type of highlighting technique is available,
sets can be defined and highlighted.
overlapping
The meanings of a color and high intensity,
for example, can be assigned to sets that are not mutually exclusive since a single
character can obviously have both highlights at once.
Highlights will be used for four types of data items in the basic sequence
display. These are: stops, segments, times and time constraints. Whole flights
will not be highlighted, since the flights selected for a particular
display have
already been chosen for their membership in some set. Furthermore, it is usually
some characteristic of the other four data items that makes a flight unique,
therefore, highlighting will be used to single out these characteristics.
Many aspects of a schedule are centered around a station, therefore, it is
especially valuable to be able to pick out particular stations from a display so that
attention may be focused on them. A scheduler may wish to focus on any of three
activities occuring at a station.
These are:
crossover activity, and flight turning.
utilization of station facilities,
Multiple color highlights can be used for
different stops as long as each color is used for only one station.
-128To observe the pattern of market service, highlights can be used on those
segments serving a given market.
Many of these segments will not actually stop
at either of the stations in the market, but nevertheless carry passengers on
through or connecting service in the markets.
Because of this, only one market
can be highlighted with color at a time, otherwise more than one color may be
assigned to the same segment.
Highlighting of times and time constraints
will be used to indicate
a
variety of relationships between different events. The most important uses of
time highlights is to indicate which times, of which events, are responsible for
making a particular slide operation illegal.
Each of these four items that could be highlighted occupy a different
location on the screen, therefore, highlighting of each can be done independently
and more than one can be highlighted simultaneously.
Stops are highlighted by
highlighting the station identifier characters and the spacing characters (dashes).
Segments are identified by the bracket characters and intervening spacing
characters
(equal signs).
Times, of course, are identified
by the two digits
representing minutes and time constraints by the constraint symbol.
-129-
4.4.3
Level Three: Combined Modifications
The three types of modification can be applied together to produce an
entirely different set of displays. Although some of these retain the formats of
earlier modified displays, it is not the combination of displays but rather the
combined use of the modification techniques that is presented here.
Combina-
tions of highlighting and scale modifications will be discussed first, followed by
combined highlighting and rearrangement
and then scale and rearrangement
Finally, highlights will be applied to this latter
combinations.
combination to
produce displays modified along all three axes.
Highlights can be applied to the 10 minute/block or 15 minute/block scales
in exactly
the same way in which they were applied to the basic format.
However, the smaller scale summarizations are highlighted somewhat differently.
In these compressed displays many of the data items that were highlighted earlier
are not present.
Thus highlights can not be applied to specific data items, rather
they will be used to identify those elements of the schedule present in the display
that contain the characteristics or events that were highlighted before.
So, for example, if a particular market is to be highlighted in a display that
only contains flight numbers, each flight that serves that market will be
highlighted.
Or, in the display of daily utilization by sequence, each sequence
that has some characteristic
on a particular day will be shown with the utilization
figure on that day highlighted. By extending highlighting in this way the scheduler
can identify those flights and sequences with certain characteristics so that they
can be isolated and displayed in a larger scale or manipulated in some way. The
-130addition of highlights to these small scale displays make them very powerful
indexes to the schedule data base.
Similarly, highlights can be applied to the rearranged displays to identify
the same information that was highlighted in the basic format. In the crossover
displays it is particularly important to identify specific markets, and in both the
crossover and turn displays, time and time constraint highlights will significantly
aid the process of sliding a flight.
In fact, this highlight is so important that it
will be produced automatically when sliding a flight produces a conflict.
Since
flashing is probably the most visible of the various highlighting techniques it will
be used to identify those times and time constraints that have reached their limits
during a slide.
When a flight is being slid, turned or crossed over, this flight or any of its
pieces should be identified in all of the displays in which it exists, so that all
relationships involving it can be observed.
As soon as one of these operations is
begun for a given flight, it will automatically be highlighted in every display that
it is part of. One highlight technique such as intensity or a single color should be
reserved for this purpose.
If blinking is reserved for conflict notification and intensity for active
working flight identification, only color or reverse field will remain for user
defined highlights.
This will still allow
non overlapping sets of data to be
highlighted one at a time. By not leaving the decision of which highlight should be
used for which purpose to the user, the task of defining a highlight will be
simplified; he only need specify what is to be highlighted, not how. The meaning
-131-
of a given highlight will be immediately apparent to anyone familiar with the
program through this standardization.
One use of highlighting in combination with rearranged displays is quite
unique. The station activity display can not be highlighted to show markets nor
more than one station; however, information on the activity at the given station
can be augmented significantly by the use of highlighting. The basic display shows
the use of the station's gates; it does not necessarily indicate the number of
on the ground at any given time.
aircraft
To do this, reverse
field or a
background color will be used as an overlay to provide a profile of the number of
aircraft on the ground throughout the period shown in the display.
Figure 4-17 shows a portion of such a display with a jagged line to indicate
the portion of the display that would be highlighted. This region extends out from
the time axis the distance corresponding to one gate for each aircraft on the
ground during the period represented by one block width.
Thus when an aircraft
departs, the line moves up toward the time axis one unit or moves out when a
flight arrives.
If an arrival and a departure occur at identical times the line
doesn't change.
However, if an arrival and departure both occur within the same
time block (5 minutes) but one or more minutes apart, the line is placed so as to
represent
time.
the minimum number of aircraft
on the ground during that block of
In this way opportunities for using under utilized aircraft or reducing the
number of aircraft required by sliding two flights can be found.
Each of the rearranged displays can be produced in modified scales that
allow more information
to be presented on one screen.
This involves smaller
scales for the two turn, two crossover and station displays as well as a larger
-132-
f- cc
C)
rS)
o-
C]
\-
kt--)
. 0...,
fY)C
f¥"i
--
CI
CO
,·G
,
L -'
I.-,
r
"
J
-T C- L-.C:
(%
v--
-
C,
LiC
L;.
Q-11
u
C; V--
r---!
C.)
r^0
"*
,-
k,
- -.L
(NJ..
i-
Cs (N
lst
CX]' (('
O
:ui
'-
C.)
E
(Ni >
=,
::-(N>t
)
.-#'
,r
-1
'o'?L
r--,,t---. ._
;:
J-S
r·
C
iL
D
~
C,']~
ti'
!-
-
. o
'--
Ci\
.:J"~C
,s,
y_
i~~~~iJLi
r',D
Lt
,_tC~
w
-...
-
(¢'
c; :-
Od
'1{D-·L_
0U
?
(d~~~~~~~~~~
r"'l
I { A W~
-- t-i?~
( C.'--s L:
C_ ·D ·--
c,
-CD
(
© '
v-- ~'
L ~~~ ~ ~.. ~- ~ ~ ~~~~~
<LD
W
..-.t!W...
'<
C'-S C') .-
:-
~~~~~~
~~~
P'LC
7
k
*
.
C:;
Ci (.J
_)
Cc
_G
J,
_
·r-J
C',
J
u3
'
,.
i...
:1h
O
-
--
a:
V
~~
-
,r-- ,-,-~n-C
01Cq(, '
-133scale for the station display. In addition, two summarized service displays that
could not be produced at all in the larger scale can be created
when both
rearrangement and scale modification techniques are applied simultaneously.
Both the 10/1 and 15/1 scales can be used readily in the two turn and two
crossover displays without any modifications.
The essence of these displays is the
relationships in time between flights and thus the compacted summaries that
exclude time are of little value with these rearrangements.
However, in each of
the rearrangements only some of the flight times are important. To compress the
turn and crossover displays requires only that those times related to the crossover station or to the turn station be retained.
Figure 4-18 shows an example of a station-time display for turns at station
A between 10:00 and 12:00.
Each line starts out with a sequence number, an
organization station, a flight number, the full time of its termination at station
AAA, and the flight number of the next flight in the sequence.
The earliest
arrival is placed on the first line, the next arrival is placed on the next line, etc.
Following each termination time is the origination time of the earliest flight that
leaves late enough to turn to the previous flight preceded by the flight number of
the actual previous flight in this sequence. Flight numbers, termination stations
and sequence numbers for the originating flight follow on each line.
Those flights which are not part of any sequence are said to be members of
a "flight pool." These flights are shown in their proper chronological sequence
with the other flights, but are marked with a "P" instead of a sequence number
and are not shown with flight numbers with which they turn. Flights that do turn
to one another can be found either by matching sequence numbers or by using the
-134-
oo0
co
oc
oLr
M
0o
lC
O
C)
L : p;
s X L[a
CO
C--
c
i
I
C
a)
000000
C-'
cL
O
cc cO O O
'C>
'-'-
a)
C-' '
-
-
c
- 00 000
CU
>-
oo
L.
O0
CO
Ln
-
O
O
OO
O OO
r)
.
Ei
L0_Z00C
Lno0
00
o'' 00
-
,
n
ko
oO .,--.......
re
O
'-0L
CUT LMMfX
c
.
000000
O
000000
m
m
N
C
'--
00 00
u
X
:2
>
z
C _) X
)
,)
L
oI
Ul
z
.r0
"O
-0
'0
CM
CO
N--
,-N C
-- LA
) C--c oaO - CMCY)
-
-
--
L '0
-
i--
t--00 C' 0'-
-
C
Co ro- Z
COJ CM
C\
C
,
-135-
flight number of the flight to which it turns. To indicate whether a flight actually
turns to the next flight in the sequence, or whether it does not because the two
flights terminate and originate at different stations, the vertical bar or slash is
placed between the time and the next flight entries for each flight.
Figure 4-19 shows a similar format used to show all of the turns made with
a given flight on each of its days of application.
Since all existing turns are
shown, the flight numbers of the flights with which each flight turns are not
shown. In their place are the days on which a given pair of turns are applicable.
Figure 4-20 is an example of a display to show all stops at station A
between 10:00 and 12:00. This provides the scheduler with the opportunity to
observe all potential crossovers in a compacted form.
Each line contains one
flight with its stop at station AAA represented by the flight number which is
centered on the line. When the stop is also a turn, both flight numbers are shown.
On either side of the line center are the arrival and departure times at the
station.
On either side of these times are the complete upline and downline
itineraries of the flight followed and preceded by the flight numbers to which this
flight turns.
At the beginning of each line is the sequence number (or a "P" for
pool flights) corresponding to the flight. The flights are arranged from top to
bottom in chronological order according to their arrival time at station A.
In Figure 4-21 a similar display is shown for all crossovers to a given flight.
The connections to this flight are shown for each of the stations in its itinerary in
separate groups, separated by a blank line, with the group for the first stop of the
flight shown first. Each group of crossovers is broken down further into groups of
-136-
os
z
ICD
U)
L
CO
I
oo
L-L)
0U l C C
\
C0n
C
..0000
CU
0000
'.0
Lco
Lf- L
Lr Ln
J
c,
LnL
- -
tOQ
af
NO
-
J0
0
,00~~
LCD
osy')
'
MO
c
0_
3-
..
~
=:
m
O
-j'
J
0
Q)s0QC
os
000
OO
Do
o
'----
'--00
O '- 0000
N.
cO
LD
- CJ M
Lr) .D
0D nr O
,-
'-
- CN
-,- --
L\lD bt-O C ' O - N m-
"r--- -
NC\jN( CjN
N
-137-
C M
Ln
0
0000
co
o003I
O0
m
Ccj
C\j
LC
L
CM
'0
P
>-L C
cz xi m
L
L)l~>~
000
CM
ULn00
C\J Ln C
0
Ca
O O0
C\M
mOO
O O0
-U
0
E
~-~jCC-
C) O
t'D
co
L
M0
00 00
0
O
0
O
OC
O0
C
co
L
0O 0O
000000
co
M
0L
U >
0
CIN
LrhLCV
O
00
Ln
= c
C)
0
CZ O >-n LT
00 -, Lr
0)
0
C.
>-, [- C Z
-0M
N
E
0
Cn
(9~1
I3I
UC'
co~
TT
0
0n
cO
0000
0000
CM CNM
M
E
a
--
T- C\J M
0000
CM
T
0
CM
L
L-0
0000 0
-
C
Mr =j· U' \O f-O O
0r-- -- -NJ- --
-
CM
CM
-)Z 0M
OCM
L-D- -- i N l NJ
-138-
000
O
J
\ aC
Uj N.,C\j
CJ
~C'j
\- xJ
O
oo
cr
tl0o
0O O
O--O
J~C
Ln
J
b-Z
M
C
oO
c-
C\0
Z
Z
0o
cJ
%DO
Cl
CO
0D
CJ
J O
CM
..
CO
T-.
...
....
C\J
r-
Ln
I
tq
+\
(f\J
(j
00
00
00
T--
=D
CU
+I
C M-
~-
q.-
'C--
C
'-
Y-
+
I c'
I +
E-
+ +
+ 100
O C
N.
I C\
0
1,lo0
IV
0
MX0
(xjv-- :
0.
+
%-,-
CM
0
LnOL
r
rM
tv- C
0+O
In
_e
m
00
e=
O
OLn
000
O
-t
CM
C
Cm +
0r
Ln OLn
00
0-- 0 v r-
q-
?CM:·
Lr.n i
L
--
.-
UJ 0o5
;
'-
o
o00
O"
-0
·
CO
C m
m
=
W=
= cQ C
0
s
E-L
E-WL
>:
or X )
C X V)
Cy X
XC
X C
X cr
CL
a.
a E-a.
zZ
Z
z
0CE
C;
.. Jls
X
= X
Hx~
=
z
M
O
--
O
0
N1
N
CO
--
0 0 00
Ln
M
O L.0
NV
O
'-
O
--
C
CU
m0 0
00
D L
CUM Z UIn'C
0
0
'
v-
LnDot-O. C. O - CMr -
(CN
0 0 .-
-
--
'-
-
-
---
v--
CM C CM CMJ
-139flights which crossover on the same set of days. These groups are separated by a
line containing the application dates for which the crossovers are valid.
In addition to the information in the previous display, this one contains
symbols to indicate whether a flight is crossed to the given flight or whether it is
crossed from it. A plus sign before the flight number indicates that passengers
can crossover from that flight to the given flight, a minus sign indicates that they
can't. Similarly, a plus a sign after the flight number indicates that the flight can
receive traffic.
Two scale modifications can be made to the station rearrangement display.
The first simply expands the scale to the largest one available: I minute/block.
This scale avoids the constraints
on minimum ground time and minimum gate
separation time that were so restrictive in the smaller scale. In this larger scale
ground times can be as short as 6 minutes, interflight times as short as 11 minutes
and gate separation need only be one minute.
This scale only permits one hour
and twenty minutes to be viewed at one time; however, many gate congestion
problems occur in this time frame.
The other way to display station activity uses a highly compressed format,
shown in Figure 4-22. Here, each operation is listed in chronological order with a
flight number, aircraft
type, arrival or departure
indicate whether it is an arrival or departure.
time, and an "A" or "D" to
If all aircraft types are included,
this display shows the density of operations at any time during the day; or, if only
one aircraft type is used, then it indicates how efficiently this aircraft type is
being turned.
-140-
0oZ
COs
a'
oo
Ln
T-
N
t.O
00
t.i
E
CMj
unC
N
0n
07
N
00
LnO
J~~~~~~~~~~~~
C"
OD
08
0'!I
LCT
x
>
lN~
zN'0-N'Nr=-
no
OD
Ln
y')
LCI
C\ 0
.>
._
Ln
_CT
T,oC0
os
-
ON
W TCD
Ln
LzL
U
rn
F
n
Ln
TC
Ln~~~~~~~~~~~~~~~~
cl
~C
0 0 C
C-
rn
O
-vt3
sO
-
t-
TC
t
-r-T
L
)-C)
rO -- -r
sO
O
\O
N
~
n -
MC 0 C) C) C C l MM
Cx CI
~
V
L;
\~~C
%o\ O
C
J
CV
N
~~Ov-v-N
fl~ ~-----ON- v-r-v-O~
v-Ln1£(In O O Lf t 0s 3 3
S O-S M
53O~
IS)-
J- Oc
MrA1pc
)-O
CEJ~L
H- ............ *
LNO
O O- J-O
_ N Mn
rC L·-t 00
-n 0- -_O
-N 1-M Ln
V- T-- - - _ --
o-
S-
J-
N N\N N
6,
E ITC
,c
-141-
Placed to the right of each entry is a count of the number of gates
occupied by a particular aircraft type following the operation described on that
line. This count will be relative to the minimum count of the day. In other words,
the lowest number of aircraft on the ground will never be less than zero.
The final scale and rearrangement
modifications will be made to produce
two displays for observing the service pattern in a given market.
A familiar OAG
format will be used to represent the service in each market for a single day. An
index will also be available to indicate which days during the period have identical
services and how many of each type of service is offered on those days.
An example of the OAG format is shown in Figure 4-23. In addition to the
departure and arrival times, aircraft type, two letter airline identifier, flight
number and number of stops, two other values are included with each service
entry to aid the scheduler. The first is the service's circuity, defined as the great
circle distance between the two stations divided by the actual distance flown.
The second number is the time efficiency, defined as the minimum time that an
average jet aircraft would take to make a direct trip, divided by the actual
duration of the service offered.
These two numbers allow the scheduler to
evaluate the quality of the services offered.
These offerings are arranged in chronological order by departure time.
Connecting services are listed somewhat differently with both flight numbers and
airline identifications and the connecting point between them, and without the
aircraft type or number of stops.
-142-
coo
L
m
H
ro
trl
0D
Ln
~.oo
aO
co
%'
Ln
o
CC)
CYN
oo
U-1
0
0
0
0
N
V)
ul
rO
<
-M
00
_-_CU
C7
m
a:
o
Cu
CD
,o
U')
I
L;4 *--OOO-CLCQ
*
*>
,- 10C
-zo, --- -r
30
3 o I_-- _o,
..
C) * . U. N M
e
Ir L. U
ai
6
,\0C[-t-- o t--
£
CM
Cvj
WO
Lmv-
J:CtX a0 v-a - oC Un C CS
¢
MDoUC
C
C\ M
-
-
y--.
CU
- CUMJ U)flD tcO CoO - cu mJ Ln Dot-_-
V,-
v-
V--
_
-
o0 - N m -CU CU Cu C
CU
-143Finally, we can highlight each of these new modified scale and rearranged
displays. Highlighting of these displays will be done in a parallel manner to the
way it was done for the other displays. Those elements present in a given display
can be highlighted, and when not present, the flight numbers that take their place
will be highlighted instead to indicate which flights contain these elements.
4.5
Interactive Commands and Working Panels
A set of displays has been described that will be generated
portions of the existing schedule.
to present
Still needed are methods for producing the data
in the schedule, manipulating it, and specifying the nature and content of these
reference displays.
A complete command language for executing these tasks is
presented here.
In keeping with the philosophy of interactive
graphics design, the com-
mands to work with the schedule will not be simply typed into the computer.
Instead, special display panels will be provided for each command that prompt the
user for required and optional data.
These displays, which will be known as
"working panels", are the only means the user will have to input information into
the computer. But, they will not merely prompt the user and accept input, they
will also respond to invalid entries with a signal or message that suggests remedial
action.
Six commands,
their accompanying
working panels, and subcommands,
provide the scheduler with complete control of the interactive graphics schedule
development process.
Each of these commands accept many data items and
-144arguments and some perform a number of related tasks.
These commands are
described individually below.
4.5.1
Access Sieve
For a number of reasons a scheduler may need to find or refer to a number
of flights or other schedule elements.
As demonstrated later, selecting and
highlighting displays and deleting flights and crossovers requires the use of a sieve
to sift through the schedule and extract those elements that meet a set of
conditions specified in the sieve. The data items that will be tested for include:
aircraft types, airlines, flight numbers, markets, stations, turns, cross-overs, and
the times of various events.
A simple and very abbreviated set of symbols and
syntax are presented to specify any conceivable combination of tests to sift
through data.
Table 4-2a contains a list of the five data items that can be used in the
specification of a sieve.
When a sequence number is used, it must be prefaced
with a vertical bar to distinguish it from a flight number. The remaining four
data items can be recognized by their identifiers through a few simple decision
rules.
Airline identifications
must be a two letter code.
Aircraft type must be
represented by a three character alphanumeric code that does not use the same
sequence of letters used by any station identifier.
A station identifier is a three
character alphabetic code. Whena three letter code is encountered it is assumed
to be a station unless it can't be found, in which case the aircraft type list is
searched.
Finally, a flight number can be from one to four alphanumeric
characters as long as it doesn't repeat any of the codes for other data items.
-145-
a)l I
I
I
I
I
II
Data item
Airline
Aircraft type
Station
Flight
Sequence
I,
1
Rules
Example
I…I
I
2 - alphabetic
[ 3 -
I
alphanumeric
, 3 - alphabetic
1-4 - alphanumeric
"" - 1-3 numeric
I
I
|
AA
DC9
AAA
101B
1,100
I
I
I
II
TABLE 4-2 a
Command Language Symbols and Syntax
-146These identifiers can be combined with the relational symbols, listed in
Table 4-2b, to produce more specific sieves.
qualified in this way.
Only stations and flights can be
Aside from these, some of the operators require some
additional information as their arguments. The greater than, less than and equal
to signs are followed by either a number of abbreviations
for the days of the
week; a date in which the month and day are separated by a semicolon; or a time,
in which the hours and minutes are separated by a colon.
As described below,
these are used in combination with other specifications to qualify the dates and
times of interest.
These three time and date qualifiers, when used alone, are assumed to
refer to all flights that originate on, before, or after the specific date or time
given. Often, these qualifiers will be used with others that qualify which events
will be tested to see if they meet the time and data specifications.
Any of the six
event symbols for cross-overs, turns, termination, origination, arrival and departure can be followed immediately by a time and/or date specification.
This usage
indicates which event is being tested by the time specification.
In addition, combinations of symbols will be used to specify another event
relationship. Either origination and termination symbols or arrival and departure
symbols, placed together, will refer to the relative duration between these two
events, rather than the absolute time of them.
So for instance the combined
symbol, (), could be used to refer to segments of a given duration.
The time
qualifier following this pair of symbols is assumed to be a duration rather than an
absolute time.
-147-
symbol t
b)
Description
,
Arguments
I
*
I
-
I
Example
~---I
I
I~~~--~~-----
I
I Flight
All services
#
1 AAA*BBB or *101
or
Pair of stations
Non stop service
I Pair of stations
I AAA-BBB or -101
I Pair of stations
+
Through service
I Flight #
or
AAA+BBB or +101
I Pair of stations
/
I
#
I
I
I I
Crossover
Termination
]
[
I
Origination
)
I
Arrival
Connecting service I Flight # or
Pair of stations
I Station
Turn
(
O
or Flight:
"
"
n
" '
' "
Departure
>
I
After
<
1
Before or -At
At
or At
TABLE 4-2b
Command Language Symbols and Syntax
or
@101
#AAA
AAA
or
or
#101
]101
I [AAA
or
[101
)AAA
or
)101
"
(AAA or
(101
'
"
rr
@AAA
"
'
,
AAA/BBB or /101
n
1 >AAA
or
>101
<AAA
1 =AAA
or
or
<101
=101
-148-
All of the qualifiers that have been discussed so far can be combined into
larger strings of qualifiers that single out even more specific parts of the
schedule.
Four symbols that will be used to aid this process are shown in Table 4-
2c. These logical operation symbols can be used to combine qualifiers into logical
statements.
Any qualifier preceded by the "not" symbol ( ) will be used to exclude the
elements identified by the qualifier rather than include them. Qualifiers can be
connected
sieves.
by either the "and" (&) or the "or"(,) symbols to form more complex
Schedule elements
must meet all of the requirements
of qualifiers
connected by "and", while they only need to meet one of those connected by "or".
The "group delimiter" (') can be used to combine a number of connected qualifiers
together by placing one at each end of the group. These groups can then be used
as independent qualifiers which themselves can be connected by logical opera-
tions.
Note that even though apostrophes don't themselves distinguish between
the beginning and end of a group, this can always be infered.
An apostrophe
preceding an operator or the end of the line closes a group and one following an
operator or the beginning of the line opens a group.
A few examples will clarify the meanings of the notations described so far.
a.
notation:
=1 2; 18'AA, TW,BN' &'DC9 ,747'
&' #AAA
>10:<11:30' $'XXX/YYY(>7:20<9:00)
>12:05<13:50
-149-
c)
Symbol
,
&
'
Meaning
,
Logical 'or'
:
Logical
Logical 'and'
'not'
, Group delimiter
TABLE 4-2c
Command Language Symbols and Syntax
-150meaning: All flights on either American, TWA or Braniff using DC9 or 747
aircraft,
that cross over at station AAA between 10:00 and 11:30 on December
first, and also serve the market from station XXX to YYY with connecting
service and with a departure between 7:20 and 9:00 and an arrival between 12:05
and 13:50.
b.
notation:
'>12; 1 < 1 27
''&'
t
&
AA&DC'
, TTW&747
&^'XXX*YYY YY*XXX'&
.A, EBBB'
^'CCC*DD(
)>1 :20,DDD*CCC(
)>1 :30'
meaning: All flights on Monday, Wednesday and Friday between December
first and seventh inclusive, that are either American DC9s or TWA 747's and that
turn at AAA or BBB, and that doesn't serve CCC to DDD with a flight time longer
than 1:20, or DDD to CCC with a flight time longer than 1:30.
Sieves are created, viewed, manipulated and deleted by using the "access
sieve" command.
When this command is entered the working panel shown in
Figure 4-24 will be produced.
Contained in this panel is a list of previously
defined sieves, each preceded by a five character name, used to refer to it in this
and other commands.
as necessary.
The list occupies two separate columns and as many pages
A sieve longer than one column wide is continued on successive
lines. When the command is first issued, the last page of sieves is shown.
-151-
0
k0
-,3
--
·
0
X
-
03
Ms
U
g:l
r
*
Z
¢*
cc
Ln
X
.
cD
_
#
1_ u
cc
U
a
f
C
Q
A-
~CC
_T
CC
cu
oo
CM
r-~~r
cO
-vO
ry)
0~~~~~~~~~
0l
U
r-
C
-A
X -/
_~1
C0
w
-
_0
oC)\
-
-
.
,
[J
C[CC)
\D
H/
-O
-O0 V *L
0
U
o
V
._
a
J
tal
:E
L)
3
C
Z
¢
v- C C-Z LI D. t- CO00
- C(M -)- L)0 t-- 0 C'0Oy-
-
r-
-
r-
*-
U
C
NCM
r:-
C
C
C
.W
C
LLC
c~~~~~
-152At the bottom of the screen on each page is a list of sub-commands, the
page number, and the total number of pages.
The commands available are:
change page, copy sieve, access sieve and delete sieve. Change page requires that
a page number be entered that does not exceed the last page of data, unless the
last page is full, in which case it can exceed it by one. Delete sieve accepts a list
of sieve names, each of which must be defined, and then removes all of these and
closes up the gaps on the screen where these entries were. Copy sieve accepts an
existing sieve name, followed by one or more new names separated by commas,
which are created with sieves identical to the old one given.
Finally, access sieve will cause the page containing the name given to be
displayed, or the last page if the name is new, and will place the cursor on the
beginning of the first line of that sieve.
Old sieves and their names can be
modified by moving the cursor, inserting new characters
and deleting old ones.
New sieves are simply typed in. Carriage returns can be entered between sieve
elements to continue the entry on more than one line.
When the data for any command has been entered it is processed by the
computer.
Sieve names are checked for legality and existence; sieves are checked
for syntax errors; and page numbers are confirmed. Errors are indicated with a
message on the bottom line of the screen and by flashing those portions of the
data entry that are illegal or inconsistence with each other. A special function
key or command will return the user to the command level from which another
command can be entered.
4.5.2
Display
-153-
The "display" command allows the user to select the type, scale, content
and highlights that a display will have. Once defined, these specifications can be
assigned to a CRT for actual implementation.
The specifications
are saved for
repeated use including deletion or modification.
When the command is invoked, a panel is shown containing the names and
specifications of every display defined with one definition per line. At the bottom
of this panel is a list of sub commands available for manipulating displays which
are:
delete display, assign display, create display, modify display, change page
and return to command level. Delete display requires one or more display names
as arguments.
Assign display requires a single display name and a number that
identifies one of the CRTs available for display and, also, the number of the page
of that display that the user wants shown. Change page allows additional pages of
display definitions to be shown by entering the page number desired.
The create-display
characters.
command requires a name consisting of up to five
Once entered, this subcommand leads the user through a series of
decisions to completely define a single display. The user is first presented with a
menu of seven types of display format, each with a selection number so that one
cand be chosen. Table 4-3 lists these options in the first column.
Depending on the format selected, the menu is replaced with from two to
four options for the display's scale.
The second column in Table 4-3
lists the
scale options in abbreviated form. The numbers refer to the number of minutes
per block, "O-T"refers to a display containing origination and termination data,
"I#" indicates that flight numbers will be shown, and "+" refers to additional
-154-
information to be displayed such as utilization, etc.
Scale options will also be
presented with numbers for the user to select.
Next, the user is presented with prompts for information
regarding the
content of the display. A number of required items are entered explicitly, then
sieves are specified to further narrow the content.
The third column of Table 4-3
lists these required data items. Prompts are issued for each individually so the
user can enter the appropriate response.
Note that for both the turn and cross-
over displays, two alternative data entries can be made: the first can be either a
station or a flight number;
the second must be a pair of times, if the first is a
station, or either an itinerary number or date of application, if the first is a flight
number.
Sieves will have slightly different
effects
when used with different
displays. The first three displays listed are sequence oriented rather than flight
oriented.
Therefore, if a flight passes the sieve, the whole sequence is included in
the display. Two sieves are allowed for the first sequence display.
One selects
those flights to be displayed with full itineraries, the other specifies those to be
shown with origination, termination
and flight number.
Flights that pass the
sieves for the turn and cross-over displays must also meet the requirements of
these displays before they will be shown.
The last two displays, do not show
complete flights, only those portions of the flights implied by the sieve, and alsi,
only those flights that have a displayable portion.
In addition, since only the cross-over and service displays can distinquish
between airlines, only these will display more than one airline's flights together.
The others
-155-
a)
·
Ha)
u- a)H
0
U)
E
4J
a)
-w
4
a)
I-
U,
·r(-4H
E-l
a)
E
a)
c
--
4-W
a)
mc
Q) F--
L)
.wI
Cd
c
H
0
C
IH
*i
rUU
4E-
C,
4-
O
0
z
~4
a)
U)
U
G)
o...
U)
4-i
CZ
a)
Q)
-
Q) -4l
-
u¢
-W
$-4
r:
C
C/)
o
I
-R-
U
+ 0
r-q-4a
la
.r4
a
T.3
U- ) -
u
*S
r--4
x
U) U a)
-Oc x
a)
--
*-4
C
C
,U~t0H
u
+
o
a
0)
ou -r.
CD
:i
a)
>::
P4
c
-156-
will arrange the flights from each airline into separate groups with the names of
the airlines separating these groups.
After valid entries for each content parameter have been received, the
computer requests a sieve to define which elements of the display are to be
highlighted.
The sieves that can be used must define a single independent set of
elements in order to be highlighted without overlapping.
Therefore, if a sieve
specifies multiple markets or multiple time bands for similar events (arrivals at
one station for instance) then this sieve can not be used for highlighting. Entry of
an invalid sieve will result in an error message and a prompt to try again.
When a sieve defines multiple conditions that must be met for a flight to
pass, each of these will be highlighted.
For example, if a flight must contain a
departure from station "A" between 10:00 and 11:00, and be flown by a DC9, then
the aircraft identifier, the station identifier and the departure time will be
highlighted. In displays where some elements are summarized, the smallest data
item that contains the selected element will be highlighted instead. So in the
above example, a daily summary display would have the aircraft
type and flight
number highlighted.
4.5.3
Access Flight
"Access Flight" is one of the most important commands available to the
scheduler using this system.
It allows entry, modification and deletion of all
flight data including operating times. The manipulation of operating times is of
course the sliding function that is so essential to schedule development.
-157The access flight command requires a flight number as an argument. Once
specified, a panel is produced with a list of the different itineraries
including
their identification number, routing, aircraft type, and dates of application. An
itinerary can then be accessed for further development or deleted by entering its
number following the appropriate subcommand. If the command to exit the
access flight mode is issued after all itineraries have been deleted, the entire
flight is erased.
A new itinerary
can be added by entering the appropriate
subcommand followed by the unique number to be assigned to this itinerary.
When a new itinerary has been accessed, a panel is produced with a number
of blank or default data fields for that itinerary and each of its segments. Figure
4-25 shows this panel with arbitrary default values where they would normally be.
Default values are defined by the user in the "global parameters"
command
described later.
This panel shows dashes where data will be entered by the user and the
"lower case x" (x) where the computer will supply information later.
The top of
the screen has information on: flight number; application days; cycle number, if
this is a defined, and sequenced flight; and equipment type, if this has been
specified. In the center of the screen is a column for the flight's routing, or
itinerary, which will be placed with one stop on every third row. To the left of
the itinerary is space for a variety of data including segment cost, fuel on board
at each departure and arrival, as well as the amount added at each stop, segment
-158-
co o
0oo
o
u~
c~
.n
r
om
X
c~
cc
f
Z
voX
Ln
X
t
XI
I
X
*e · ·
:E~
CU
X
X
X
X
X
X
X
X
X
X
X
I
I oI
I 0
··
I
......................................
--
I o
I
I o
··
i I 0
··
O
I
I
CI:E
I
i0
0
I
I
I
I
I
I
I
I
I
i
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
o
-
,U U
Ln
r~
X
X
X
i
i0I
I
I
I 0
0
Z
C
s
X
I o
0I
cO
0'
X
XXXXXXXXXXXXXXXXXX
LL
c~
X
X
X X X
X X
X XX XXXX
XXX XXXX XXX
XXX XX XXXX XX XX
I1
·· · ·· · · ·
...........
I..........
I
I
I
I
I
I
I
I
I
I
I
I
II!
I
I
I O
I CI
I
I
I O
I CI
I
I
I O
I CI
I
I
I O
I C
I
I
I......
I
I
I
I
I
I
I
I
I
I
I
I
I
a
I O
I CN
I
I
I i.
I
m
I....
O
(~I
-X
~-
k
Ch Ln0
OZZH
H :
c0
P_
\0
U)
II
· 0· ·
I · I°
I I 0
I ·1I °
i I 0
~~~
EH I X
X4I
I
X
x
~~~~
X
x
x
I
X
X
x
"
=
1I IO0°
I IO
IX
x
X
E>
_
x
o:~
UH 2
x
X
X
C_)
X
0o
C\1
I S
x
Lt
I
I
IX
I
c
( x¢
x
x
x
x
x
x
X
X
x
X
X
x
X
X
x
X
X
x
X X
X
X
X
X
X
X
X
LL
..
*-
-
I =
:
I-
....
E
C4
X X X X X X X X X X X X X X X X X X X
00
{kD
cO
CYO
n
U
=
ct)
..
CU
+
II II + II II +
X
x
X
X
X
X
X x
X x
X x
X x
X x
X xX x
X x
X x
X X
II II + II II + II II + i II +
<
x
X
x
x x
X
x
X
X
X
X
O
x
x
X
X
=:J
x
X
x
x
X
X
X
x
0
¢
0
OD
CT
Ln
J
M
CY)
a~C
C3
.z
H
¢Q
j
* N m T m\n
Foot
:r:
- Um M
o\
v-
v-
r-
CtDCo
LRFO
\C
-
-
-
t -
t
NU N
N M U
N
U~~~~~~~~~~~~~~
c
C
.
~J ,L xX xX x
X x
X x
X x
X x
X x
X
C
M
1
N
CC.
_Z
I ~I x x x x x x x x x x x x x x x x x x x
r~
L.
-Y
V)
:D
co~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~,.
L()
tL
tN
C
*
H
!
IX X
I
\.o x x(x ( x (
~
CO
1I· 1I °
I I
XXXXXXXXXXXXXXXX
~J_
X~~ I xX* xXXXX
xxx x xX xXxXxxXxX xXx xXxXx xXXx xXXx
X
m
CO
0·°1
· ·1· I I° ·
I 0
I I O
X X X X X X X X X X X X X X X X X X
X X X X X X X X X X X X X X X X X X
'5
UZ
Cr7
an~
CD 00
~
I
I
N
.
-159distance and flight time.*
These items will all be completed automatically when
the flight is entered.
To the right of the itinerary list are five columns containing times that
relate to the flight's operation. On the lines containing station identifier, the
times relate to ground time at that station. Preceding lines contain arrival times
while lines after the station lines show departure times. The third column of each
line of five columes is the actual departure, arrival or ground time.
On either
side of this is the minimum and maximum value that the actual, or nominal, time
may be as specified explictly by the user. Finally, the two outside time values are
generated automatically by the computer when the minimum or maximum times
permitted without breaking connections or turns are more constraining than the
user specified times. Note that global default values are placed in the minimum
and maximum ground time locations for each station.
In addition to the constraint time values, each constraint is also represented by the symbol representing its type (pinned, bracketed, limited or
unconstrained). Constraint symbols are placed immediately under each station
identifier for the operating times constraint, and immediately to the right of the
station identifier for ground time constraints.
specified the defaults are:
Before any constraints are
unconstrained operating times (underscore) and
bracketed ground times (caret).
The only subcommands needed to use the itinerary access facility are:
change page, enter data, exclude services, and return to access flight command
level. All other actions that need to be taken can be controlled by moving the
*Other data might be included such as estimated traffic or revenue, but these will
not be discussed here.
-160data.
cursor to the proper location and entering
movements will be restricted
To aid the user, cursor
to those locations where data can be entered and
even further restricted to that subset of locations that can be entered given the
current state of the entry process.
Thus, if one data item must be entered before
another or can't be entered if another already has been, the cursor movement will
be restricted to currently legal locations only.
When a new itinerary is being entered the cursor will begin at the aircraft
type location.
An aircraft type identifier can be typed in or deleted and retyped.
A down-cursor will cause the aircraft type to be entered and validated.
Once a
valid entry is received, the cursor will move to the first station identifier
location.
A station identifier
is entered
on every third line by moving to
successive locations with the down-cursor. When a right-cursor is typed the
entire itinerary will be validated and any invalid stations will blink.
Cursor
movement remains restricted to itineraries until each station entered is valid.
When a valid itinerary has been received the computer will return values for each
of the data items on the left side of the screen including the flight duration times.
Figure 4-26 shows a typical completed access flight working panel.
After a valid itinerary has been recorded, the cursor will move to the first
nominal time location, which is the departure time from the first station.
The
cursor is then restricted to movement between this departure, the adjacent
arrival and the following ground time until any one of these three times has been
entered.
As soon as one of these three times has been entered the cursor will
move to the first of another set of three times. If either an arrival or departure
was entered, the cursor will be restricted the same ground time (the ground time
at the second stop) and the next pair of departure and arrival times. If, instead, a
-161-
Coo
cr
NJ
J
O
O O O OM
X
¢ 0:.....:: ..00
. O
.
0G
cO
cc)
' X,- 0 O- I.)
L: O
OCD
'.
Cj
O
\D
~
b--
O
o o
o oC5
%
-l
i
O
Ll
O0
MtC
)
)J0
: 00
OH
M:
C
P:
w
B
c
C)
O
O
n
U
U
0H H
0
0
O
O
--
\JO
¢
c
'C)
O",
Com
o
~
<
=
O
EH ko
LO
O
C
0o
J
H
2
C
0
\0
x
'
x
0
LI
-J
c
-
..
.
II II + II II
0
\'0
0 U
T-
D
O
.
+
ULLU
X
E¢: 0
C' Z
~i~f
L2"
<
'
L
njU
X.0..
0n
UJ
\.D
(NJ
H
O
000000
co
0
U
U
O4
.'O
rd
r,
n
O
X
Cn~
.)
......
oo
O O O O,'-C-
O->
¢J
M
"
r~ rO
r~.--
.
CI
rN
0c
'--'
[.,
IU
ZO
=
'.\o v-- m CToC
- CNj - uI'o\D-- T- - T- rYr
r-
a'o
T
N
-
CMN
M
-
N N
-162-
ground time was entered first, the cursor will move to the departure, arrival and
ground times immediately following the ground time entered.
This same pattern continues with the currently accessed group of three
times beginning with a departure time until an arrival or departure is entered and
then beginning with a ground time until the last group of three.
includes the final ground time, which is the inter-flight
If this last group
time, the cursor will be
restricted to only the departure and arrival time from this group. This is because
it is the departure time of the flight and the origination time of the flight that
this one is eventually turned to that will determine this time.
At any point in this process the entry in the current group of three can be
deleted.
If this is done the cursor can be moved, by an up-cursor key stroke, to
the previous group of three where the entry can be modified or deleted to permit
access to the previous group. When the last time is entered and a down-cursor is
keyed, the complete set of flight times is entered. The computer responds by
filling in all of the blank times that were not explicitly entered.
The cursor then
moves to the first minimum time constraint.
From this point the cursor can move from a minimum to the next minimum
below it with a down-cursor, or to the maximum on the same line with a right
cursor.
Up-cursor and left-cursor reverse these movements, and right-cursor in a
maximum position places the cursor at the minimum location on the next line
(vice versa for the left cursor). Any number of these constraints can be entered,
regardless of which times were initially specified.
When a down cursor is typed
while the cursor is on the last line, the complete set of constraints is entered into
the computer.
The program then computes and returns the least constraining
-163consistent set possible. Thus if a minimum ground time is set to be thirty minutes
but the maximum arrival and minimum departure surrounding it are only twenty
minutes apart, then the ground time constraint is changed to twenty minutes. In
addition, all implicit constraints are also returned and the constraint-type symbols
are set to their proper values.
If desired, the "exclude services" command can be invoked. This produces
a panel with a list of all possible markets that could be served with this flight.
They are arranged with all markets with the same origin together and ordered
chronologically. After each market is the great circle distance, actual distance,
and the ratio of these as well as the same three numbers for trip times. To the
left of each market is the word yes, indicating that this service if offered. The
cursor can move up and down this column using the up and down-cursor.
To
change one yes to a no, the right or left cursor is typed while the cursor is on a
line.
This flips the indicator to its opposite and moves the cursor up for left or
down for right.
Once a valid aircraft,
routing, set of times and constraints
have been
entered, the flight itinerary becomes part of the schedule data base. If the return
to access-flight command is issued at any point before this, the data entered so
far will be deleted. After the flight is completely entered, it can only be deleted
by the "delete flight" command. Also at this time, the new flight is displayed on
any screen which has a sieve that allows the flight to pass.
Any further entry of data for this flight will be considered a modification
which is governed by a separate set of rules and cursor movements.
If, before any
modifications are made, the flight is crossed or turned with other flights, the
-164-
display will reflect this fact by including the more constrained set of time
constraints and the sequence number.
Once a complete itinerary
is accessed the system is in "modify mode"
which allows for more cursor freedom.
The cursor can now go from the aircraft
type to the first station and from there to successive stations or to the nominal
time or either explicit time constraints.
constraint
Between the station and first time
is a vertical bar to which the cursor can also move.
This will be
explained below.
Four kinds of modification
are possible.
One, modifying the market
exclusions, is performed with the market exclusion command and does not involve
times at all. The other three, modifying the aircraft type, the itinerary and the
operating times, all affect operating times in one way or another.
Each of these
modifications are made by moving the cursor to the proper location, typing in the
new value and then moving the cursor away from that location.
Before any of these changes that might require a change of operating times
is made, any of the vertical bars to the right of the station can be changed to
dashes.
If a dash is placed next to an arrival, departure,
or ground time, this
time is temporarily frozen and will not change during a modification.
By setting
various combinations of dashes and vertical bars, the way in which times will
change can be altered.
For example, if a string of adjacent ground times are
frozen then a flight can slide earlier or later, but can not be compressed in this
region. Freezing adjacent arrivals and departures, while permitted, is redundant
since one determines the other in combination with a fixed flight time.
-165An operating time change will be accepted by the computer which will then
try to implement it.
To do this, it will start at the changed time and start
pushing or pulling adjacent
permit.
times as far as the most limiting constraint
will
If a ground time is changed, it will push or pull in both directions around
it in equal amounts.
If this sliding process encounters
a constraint
which is
insurpassable, the process stops, the times shown become the partially slid values,
and the data items representing the responsible constraints are flashed, both on
this panel and in every display in which these constraints are present.
A conflict
occurs when an absolute, fixed operating time limit is reached and all intervening
ground times are at their limits. Therefore, all of these contributing constraints
are flashed.
At this point the cursor can only move between the modified time and the
responsible constraint
values, if they are internal ones.
Once a constraint
is
modified the slide continues until another conflict is encountered or until the slide
is successful. If instead of correcting the problem, the return to access flight
level is requested, the original times are restored and the slide is discontinued.
Modifyingthe aircraft type will often require substantial readjustments of
operating times since a different crusing speed will change all of the flight times.
If this is the case then a complex sliding process will begin automatically.
Each
segment is expanded or compressed one at a time, beginning with the center if
there are an odd number of segments, or just to the early side of center if there
are an even number, and then moving outward one by one. Segments are expanded
or compressed equal amounts on both ends (arrival and departure),
constraint prohibits this.
unless a
Arrival and departure times are also changed one at a
time so that if a constraint is reached, adjacent times can be modified to permit
-166-
the changing segment to change. It is like a series of individual slides driven by a
different time change each time.
As soon as a conflict is reached
constraints are flashed.
the process stops and the responsible
In addition, all of the flight times of as yet unmodifed
segments also begin to flash.
The cursor will again only move between the
aircraft type and the constraint values responsible for the conflict until either a
constraint is relaxed, the slide is terminated by restoring the aircraft type to its
former value or by returning to the access flight command level.
Routing modifications are the most difficult since they may require
substantial shifts in scheduled times at other stations than the one modified
causing many turns and connections to be broken. To maintain a degree of order
in this process, routing modifications will be limited to changing one stop at a
time so that adjustments can be made in a logical manner. An existing stop can
be deleted, a stop can be changed to a different station or a gap can be placed
between two stops and then filled in with a new station.
A deletion is accomplished by moving the cursor to the station identifier
and hitting the delete key once for each letter.
When the cursor is moved away
from this location the stop is eliminated and all connections to it are dropped. If
it was the first or last stop in the flight, the turn to the next flight is eliminated
by the flight remains assigned to the same sequence.
At this point, the gap created by the missing stop is closed up. Since the
block time between the two stations surrounding the one eliminated will probably
be different than the two block times and ground time previously between them,
-167the operating times and ground times surrounding the gap must be slid. This is
done by attempting to slide each side of the flight equal amounts until either the
gap is closed or a constraint is reached.
As before, constraints
will blink and
cursor movement will permit them to be changed or the itinerary to be restored
to its original value.
Step changes are handled in a similar manner, except that the flight
operating times and constraints set for the former stop will be modified or
eliminated. If the slide required to fit the new station is successful, the new
arrival and departure times at this station are set, and the ground time is set at
the value that requires the smallest changes in other times and still meets the
global limits. Time constraints on this station are extrapolated from constraints
on adjacent stations. If any of these times are unsatisfactory, further modifications can be made.
Stops can be inserted by typing three blanks over the location in the flight
where the insertion is desired, this will cause the stop at that location to be
shifted down on the screen along with its other information.
A blank line will thus
be available to type in a new stop. The sliding process then takes place to fit this
new portion of the flight in between the two stops on either side. Once again, if
time constraints do not permit this action to be taken, they must be modified
before the new stop will become valid.
Upon entering a station, the computer will check it for validity according
to other criteria.
It must be a valid station, compatible with the aircraft type and
within range of the preceding and following stations. If any of these conditions
are not met, the identifier will blink until a correction is made.
-1684.5.4
Sequence
The sequence command will generate the display shown in Figure 4-27.
The date must always be filled in. One of the blanks, in each of the two columns
(either for a flight number or sequence number)
must be filled in.
A flight
number and sequence number combination will assign the given flight from the
flight pool to the given sequence.
The reverse will remove the flight from the
sequence. If two sequence numbers are specified, the entire contents of these
two sequences will be swapped. This amounts to renumbering the sequences. The
combination of two flight numbers will cause their locations to be exchanged.
The program will stay in sequence mode until the command to return to the
previous level of the command structure is received.
Any incomplete entries will
be deleted if this is done. Also, if a flight cannot fit into a sequence because its
aircraft type operating times are incompatible, then the invalid entries will blink.
If left unmodified when exiting this command, the invalid entries will be deleted.
4.5.5
Crossover
When issued, the access crossover command will produce a display con-
taining blanks for two flight numbers of station identifier and a date.
Cursor
movement will be restricted to these locations until they are all entered and
determined to be valid. Valid entries are flights that both operate on the given
date (and possibly others as well) and stop at the given station.
will blink until they are corrected.
Invalid entries
-169-
00
o
co
tCM
00
m
t0
3
t--
0
Z
z
03'l
ZT
_
IC
I
00
L:
-..
C)
T'
0)
.o
I:N
·rct
LI..
LT
.
CO
O
Lno-
_
M
L
-
C\"rM,
L
O
. t-
C
C-, O C
C\J
-M U) Of
,- ,I- ,-.. - -
--C C 0 cm- C M M
,r- - - ,_ C 0N, d ..
-170Once a valid set of specifications
has been received, the computer will
display the global minimum and maximumcrossover times next to the information
specified. Between these will be the actual time available for crossovers, which
may have a negative value. A second column with the same information will also
be produced,
but with the flight numbers in reverse order.
The two columns
correspond to crossovers from one flight to the other and vice versa.
Under each
heading is a list of markets that could be served by a crossover between these
flights.
If the global constraints do not permit a crossover to occur, then the
binding constraint will blink until the cursor is moved to this location and the
constraint is changed.
No markets will be displayed until the crossover becomes
legal. When they are displayed, the markets will also have a circuity value beside
than to assist the scheduler in determining the benefits to be derived by including
the service. Also beside each market is the letter Y or N, for yes or no indicating
whether or not the service is to be offered.
Initially all of these will be set to
yes. To exclude a service, the cursor must be moved vertically to the apprpriate
time and then either a left or right cursor will flip the indication.
to the Flight service exclusion facility described earlier.
This is similar
If more than one page of
markets are required, a change page command can be issued along with a page
number.
If a crossover that has already been created with this command is accessed
again, the entire display is reproduced.
Any of the markets can then be included
or excluded as desired. To delete the entire crossover, a subcommand can be
issued which will erase it from the computer memory.
-171-
4.5.6
Access Global Parameters
Many of the default data values that are used in various operations can be
defined with the access global parameters command.
associated with individual stations.
All of these values will be
Thus the panel produced under this command
will be a list of all stations in alphabetical order followed by their identifier codes
and then a set of parameters.
At the top of this list will be a location to enter a
single number for each parameter which will apply to all stations not otherwise
specif ied.
The parameters
to be determined are:
minimum and maximum ground
time, minimum and maximum gate spacing time and minimum and maximum
crossover-time.
These numbers are entered by moving the cursor to the proper
location and typing them in.
Values can be deleted in a similar manner.
locations left blank will take on the value in the first row of the display.
Any
-172-
5.
CONCLUSIONS AND RECOMMENDATIONS
The goal of this study was to present a design for an interactive graphics
tool that would encompass the full complexity of the scheduling task so that it
might play a central role in the development process.
To achieve this it has been
demonstrated that careful attention must be paid to the data access and
manipulation requirements of the scheduler.
The operations of -turning, crossing
over, and sliding flights have been identified as the fundamental functions of the
scheduling process.
Displays and commands have been designed around these
requirements.
Although not widely used as a manual graphics aid, the sequence chart
(often called a cycle chart) was determined
implementation
representation
to have the best structure
with respect to a CRT type terminal.
for
This was due to its visual
of time inforrrmation and untangled format, essential for displays
with a finite number of addressable locations.
To provide a complete package of
displays for presenting the unique information required for several different tasks,
this basic display was modified in scale, rearranged and highlighted.
The interactive portion of the design encompasses a set of working panels
on which the user can enter data and receive responses concerning the validity
and impact of the input data.
Restricted
cursor mrnovement, highlighting and
prompting messages are all used to guide the user through the data entry process.
A powerful and concise command language is provided to sift through the schedule
data base and identify those elements that meet the specified criteria so that
they may be displayed or highlighted. This capability allows displays to be easily
-173customized for an unlimited number of different purposes -- a major advantage of
a computerized tool over manual ones.
It was determined that the computer should not take over any decision-
making responsibilities from the scheduler. Instead, information that is likely to
be of value during a particular operation is efficiently presented for evaluation,
by the scheduler. When a flight is being slid to align events in time, the computer
indicates potential problems such as excessively short ground times or broken
connections; but it is the scheduler who resolves these problems.
In general, the computer tools described here are expected to be welcomed
by schedulers whose critical
and difficult
job has long been the subject of
automation attempts, but has only recently been recognized as an essential
element in any mechanized system.
The power of a computer combined with
human expertise results in the possibility of more and improved schedules during
each planning period, offering airlines more options from which to choose in
dealing with the pressures of increased competition and rising costs.
It is recommended that this design be implemented on a dedicated, stand-
alone computer system. For interactive graphics to be effective the computer
must keep up with the thought processes of the user (indicating the need for
response time under one second-a requirement which can rarely be met by time
sharing systems.)
It is also advantageous to keep full data base and operating
control accessible to the users. The computer should be linked to three or more
terminals arranged in a work station to permit one user to view several displays
-174-
relevant to a given decision simultaneously while interacting
with the single
"command" terminal.
Further development of this system should proceed along two lines: First,
the displays may require more customization for use by individual airlines to meet
unique requirements.
The current work provides a general tool to be adapted to
users' needs. This work should also extend to the design of displays and summary
reports more applicable to hard copy format.
Since many airlines already produce
this kind of output, these designs are expected to be highly specialized.
The second area that needs to be developed is the integration of the system
presented here with other schedule development and evalution tools. Access to a
large data base of historical schedules and traffic, cost and revenue statistics for
the user's and competing airlines would be a valuable addition.
Forecasting and
analysis tools that can predict the impact of schedule modifications should also be
included.
By placing the scheduler at a computer terminal this system will increase
the attractiveness
of these other packages that can provide immediate online
access to valuable information.
In this way, the role of operations research and
computerized scheduling tools can begin to play the role in assisting schedulers of
which they are capable.
-175-
BIBLIOGRAPHY
Akel, O.J.
Dynamic Scheduling in Airline Operations.
M.I.T. Flight
Transportation Laboratory Report, R-67-1, December, 1967.
Elias, Antonio L. The Development of an Operational Game for the U.S.
Domestic Airline Industry. M.I.T. Flight Transportation Laboratory, R67-1, February, 1979.
Elias, A., Simpson, R.W., Taneja, N.K., and Austrotas, R.A. New Directions
Presented at First World Airline
in Airline Industry Automation.
Conference, Phoenix, Arizona, February, 1979.
Elias, A., Competitive Airline Strategy Simulation: Participant File Processor User's Guide, Flight Transportation Associates Report, November
1979.
Etschmaier, M. Schedules Construction and Evaluation for Short and Medium
Range Corporate Planning. Proceedings of the Tenth AGIFORS Symposium, 1970.
Frey, Kelly L. and Ponder, Ronny 3. "New Scheduling Developments at
Federal Express." Air Transportation Conference.
Girard, Denis. The Airline Operations Model: A Schedule Development and
Evaluation Tool. Proceedings of the Thirteenth AGIFORS Symposium,
September, 1973.
Gleen, C.H. "Factors to be Considered in Airline Scheduling."
Aeronautics and Space Journal (June, 1972).
Canadian
Labombarda, P. and Nicoletti, B. Aircraft Rotations by Computer.
ceedings of the Eleventh AGIFORS Symposium, October, 1971.
Pro-
Labovitz, David E. and Jorgensen, Michael V. A System for Competitive
Analysis of Airline Schedules. Potomac Scheduling Report, 1978.
Larson, Robert E. A Dynamic Programming Approach to Airline Scheduling.
Proceedings of the AGIFORS Symposium.
Loughran, Brian P. An Airline Schedule Construction Model. Proceedings of
the Twelfth AGIFORS Symposium, September, 1972.
Mathaisel, Dennis F.X. Airline Schedule Construction Model. M.S. Thesis,
University of California Irvine, May, 1975.
Niedere, M., Frequency and Schedule Optimization by LP Proceedings of the
Eleventh AGIFORS Symposium, October 1971.
-176Pollack, Maurice. "An Interactive Approach to Airline Route-Frequency
Planning." Information Processing (1974).
Simpson, Robert W.
Computerized
Schedule Construction
for an Airbus
Transportation System. M.I.T. Flight Transportation Laboratory Report,
FT-66-3, December, 1966.
Simpson, Robert W.
Scheduling and Routing Models for Airline Systems.
M.I.T. Flight Transportation Laboratory Report, R683, December, 1969.
Smith, Charlyne M. and Kyle, Jon S. On-line Flight Schedule Development.
Proceedings of the Twelfth AGIFORS Symposium, December, 1972.
Speyer, Jean-Jacques
M.I.T., September,
W. Planning Processes in U.S. Airlines.
M.S. Thesis,
1978.
Taneja, Nawal K., The U.S. Airfreight Industry, Lexington Books, Lexington,
Mass., 1979.
Tipton, Stuart G., et al. Airline Scheduling: Management for Service.
Transport Association of America Report, 1961.
Useros, MaTeresa.
Version System.
October, 1975.
Air
Flight Schedule Mechanization Project Frequencies
Proceedings of the Fifteenth AGIFORS Symposium,
Walker-Powell, A.J. Aircraft Scheduling by Computer.
Ninth AGIFORS Symposium, October, 1969.
Proceedings of the
Williamson, W.G. Computer Programs for Fleet and Schedule Planning.
Proceedings of the Seventh AGIFORS Symposium, October, 1967.